Dunn_Marina_Assignment4.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"Marina Dunn\n\nPHYS 243 HW4\n\nFall 2021\n\nProf. Mobasher\n\nGoal: Find all the 9s! Use a decision tree algorithm on the MNIST dataset here (https://www.kaggle.com/c/digit-recognizer/data) to find all of the 9s. Compare the performance with a random forest algorithm.","metadata":{}},{"cell_type":"code","source":"import pandas as pd\nimport numpy as np\nfrom sklearn.linear_model import Perceptron\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import confusion_matrix\nimport math\nfrom collections import Counter\nimport seaborn as sns\nimport matplotlib.pyplot as plt \nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.tree import DecisionTreeClassifier\nfrom sklearn.metrics import multilabel_confusion_matrix\nfrom statistics import mean\nfrom sklearn.metrics import accuracy_score\nfrom sklearn.model_selection import cross_val_score\nfrom sklearn import metrics\nimport warnings\nwarnings.filterwarnings('ignore')\n","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2021-12-06T21:54:19.976182Z","iopub.execute_input":"2021-12-06T21:54:19.976848Z","iopub.status.idle":"2021-12-06T21:54:20.565233Z","shell.execute_reply.started":"2021-12-06T21:54:19.976775Z","shell.execute_reply":"2021-12-06T21:54:20.564345Z"},"trusted":true},"execution_count":1,"outputs":[]},{"cell_type":"code","source":"train = pd.read_csv(\"../input/digit-recognizer/train.csv\",header=0)\ntest = pd.read_csv(\"../input/digit-recognizer/test.csv\",header=0) \nsample = pd.read_csv(\"../input/digit-recognizer/sample_submission.csv\",header=0)\n\ntrain.head(10) ","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:54:20.570755Z","iopub.execute_input":"2021-12-06T21:54:20.571013Z","iopub.status.idle":"2021-12-06T21:54:25.873870Z","shell.execute_reply.started":"2021-12-06T21:54:20.570981Z","shell.execute_reply":"2021-12-06T21:54:25.872833Z"},"trusted":true},"execution_count":2,"outputs":[{"execution_count":2,"output_type":"execute_result","data":{"text/plain":"   label  pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  \\\n0      1       0       0       0       0       0       0       0       0   \n1      0       0       0       0       0       0       0       0       0   \n2      1       0       0       0       0       0       0       0       0   \n3      4       0       0       0       0       0       0       0       0   \n4      0       0       0       0       0       0       0       0       0   \n5      0       0       0       0       0       0       0       0       0   \n6      7       0       0       0       0       0       0       0       0   \n7      3       0       0       0       0       0       0       0       0   \n8      5       0       0       0       0       0       0       0       0   \n9      3       0       0       0       0       0       0       0       0   \n\n   pixel8  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n0       0  ...         0         0         0         0         0         0   \n1       0  ...         0         0         0         0         0         0   \n2       0  ...         0         0         0         0         0         0   \n3       0  ...         0         0         0         0         0         0   \n4       0  ...         0         0         0         0         0         0   \n5       0  ...         0         0         0         0         0         0   \n6       0  ...         0         0         0         0         0         0   \n7       0  ...         0         0         0         0         0         0   \n8       0  ...         0         0         0         0         0         0   \n9       0  ...         0         0         0         0         0         0   \n\n   pixel780  pixel781  pixel782  pixel783  \n0         0         0         0         0  \n1         0         0         0         0  \n2         0         0         0         0  \n3         0         0         0         0  \n4         0         0         0         0  \n5         0         0         0         0  \n6         0         0         0         0  \n7         0         0         0         0  \n8         0         0         0         0  \n9         0         0         0         0  \n\n[10 rows x 785 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>label</th>\n      <th>pixel0</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>...</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>4</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>7</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>5</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>10 rows × 785 columns</p>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"test.head(10)","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:54:25.875556Z","iopub.execute_input":"2021-12-06T21:54:25.876120Z","iopub.status.idle":"2021-12-06T21:54:25.899460Z","shell.execute_reply.started":"2021-12-06T21:54:25.876079Z","shell.execute_reply":"2021-12-06T21:54:25.898805Z"},"trusted":true},"execution_count":3,"outputs":[{"execution_count":3,"output_type":"execute_result","data":{"text/plain":"   pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  pixel8  \\\n0       0       0       0       0       0       0       0       0       0   \n1       0       0       0       0       0       0       0       0       0   \n2       0       0       0       0       0       0       0       0       0   \n3       0       0       0       0       0       0       0       0       0   \n4       0       0       0       0       0       0       0       0       0   \n5       0       0       0       0       0       0       0       0       0   \n6       0       0       0       0       0       0       0       0       0   \n7       0       0       0       0       0       0       0       0       0   \n8       0       0       0       0       0       0       0       0       0   \n9       0       0       0       0       0       0       0       0       0   \n\n   pixel9  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n0       0  ...         0         0         0         0         0         0   \n1       0  ...         0         0         0         0         0         0   \n2       0  ...         0         0         0         0         0         0   \n3       0  ...         0         0         0         0         0         0   \n4       0  ...         0         0         0         0         0         0   \n5       0  ...         0         0         0         0         0         0   \n6       0  ...         0         0         0         0         0         0   \n7       0  ...         0         0         0         0         0         0   \n8       0  ...         0         0         0         0         0         0   \n9       0  ...         0         0         0         0         0         0   \n\n   pixel780  pixel781  pixel782  pixel783  \n0         0         0         0         0  \n1         0         0         0         0  \n2         0         0         0         0  \n3         0         0         0         0  \n4         0         0         0         0  \n5         0         0         0         0  \n6         0         0         0         0  \n7         0         0         0         0  \n8         0         0         0         0  \n9         0         0         0         0  \n\n[10 rows x 784 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>pixel0</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>pixel9</th>\n      <th>...</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>10 rows × 784 columns</p>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"# Check data\ntrain.isnull().any().describe()","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:54:25.901315Z","iopub.execute_input":"2021-12-06T21:54:25.902062Z","iopub.status.idle":"2021-12-06T21:54:25.924593Z","shell.execute_reply.started":"2021-12-06T21:54:25.902027Z","shell.execute_reply":"2021-12-06T21:54:25.923527Z"},"trusted":true},"execution_count":4,"outputs":[{"execution_count":4,"output_type":"execute_result","data":{"text/plain":"count       785\nunique        1\ntop       False\nfreq        785\ndtype: object"},"metadata":{}}]},{"cell_type":"code","source":"test.isnull().any().describe()","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:54:25.925780Z","iopub.execute_input":"2021-12-06T21:54:25.926383Z","iopub.status.idle":"2021-12-06T21:54:25.946332Z","shell.execute_reply.started":"2021-12-06T21:54:25.926325Z","shell.execute_reply":"2021-12-06T21:54:25.945642Z"},"trusted":true},"execution_count":5,"outputs":[{"execution_count":5,"output_type":"execute_result","data":{"text/plain":"count       784\nunique        1\ntop       False\nfreq        784\ndtype: object"},"metadata":{}}]},{"cell_type":"code","source":"dst_train = Counter(train['label'])\nprint(dst_train)\n\n# Plot distribution\nsns.countplot(train['label'])","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:54:25.947704Z","iopub.execute_input":"2021-12-06T21:54:25.947939Z","iopub.status.idle":"2021-12-06T21:54:26.242956Z","shell.execute_reply.started":"2021-12-06T21:54:25.947910Z","shell.execute_reply":"2021-12-06T21:54:26.241662Z"},"trusted":true},"execution_count":6,"outputs":[{"name":"stdout","text":"Counter({1: 4684, 7: 4401, 3: 4351, 9: 4188, 2: 4177, 6: 4137, 0: 4132, 4: 4072, 8: 4063, 5: 3795})\n","output_type":"stream"},{"execution_count":6,"output_type":"execute_result","data":{"text/plain":"<AxesSubplot:xlabel='label', ylabel='count'>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"X = train.iloc[:, 1:]\ny = train['label'].tolist() # target values\n\n# Select 10000 rows data as a testing dataset\nX_test = X.iloc[0:10000, :].values.astype('float32') # all pixel values \ny_test = y[0:10000] # Select label for testing data\nX_train = X.iloc[10000:, :].values.astype('float32') # all pixel values \ny_train = y[10000:]","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:54:26.244368Z","iopub.execute_input":"2021-12-06T21:54:26.244670Z","iopub.status.idle":"2021-12-06T21:54:26.312631Z","shell.execute_reply.started":"2021-12-06T21:54:26.244632Z","shell.execute_reply":"2021-12-06T21:54:26.311668Z"},"trusted":true},"execution_count":7,"outputs":[]},{"cell_type":"code","source":"print((min(X_train[2]), max(X_train[2])))\n\n# Normalizing the data, pixel intensities are currently between the range of 0 and 255\nX_train = X_train/255.0\nX_test = X_test/255.0\n\n# Printing the shape of the Datasets\nprint('X_train shape:', X_train.shape)\nprint(X_train.shape[0], 'train samples')\nprint(X_test.shape[0], 'test samples')","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:54:26.313969Z","iopub.execute_input":"2021-12-06T21:54:26.314313Z","iopub.status.idle":"2021-12-06T21:54:26.373593Z","shell.execute_reply.started":"2021-12-06T21:54:26.314269Z","shell.execute_reply":"2021-12-06T21:54:26.372468Z"},"trusted":true},"execution_count":8,"outputs":[{"name":"stdout","text":"(0.0, 255.0)\nX_train shape: (32000, 784)\n32000 train samples\n10000 test samples\n","output_type":"stream"}]},{"cell_type":"markdown","source":"We find the score of Decision Tree first, then Random Forest, and for both we will use both to predict where the label of the image is equal to '9'.","metadata":{}},{"cell_type":"code","source":"# use decision tree algorithm first\n\nfrom sklearn import tree\nfrom sklearn.model_selection import GridSearchCV\n\nmodel = DecisionTreeClassifier()\nparameters = {'max_features': ('auto', 'sqrt', 'log2'), 'max_depth':[1,100], 'min_samples_split':[2,100], 'min_samples_leaf': [1,100], 'min_weight_fraction_leaf': [0,0.5], 'random_state': [0,100], 'max_leaf_nodes':[0,100]}\nclf = GridSearchCV(model, parameters)\nclf.fit(X_train, y_train)\nclf.score(X_train, y_train) * 100","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:54:26.375108Z","iopub.execute_input":"2021-12-06T21:54:26.375459Z","iopub.status.idle":"2021-12-06T21:57:01.766901Z","shell.execute_reply.started":"2021-12-06T21:54:26.375411Z","shell.execute_reply":"2021-12-06T21:57:01.766109Z"},"trusted":true},"execution_count":9,"outputs":[{"execution_count":9,"output_type":"execute_result","data":{"text/plain":"69.78125"},"metadata":{}}]},{"cell_type":"code","source":"clf.best_params_","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:57:01.767978Z","iopub.execute_input":"2021-12-06T21:57:01.768696Z","iopub.status.idle":"2021-12-06T21:57:01.775802Z","shell.execute_reply.started":"2021-12-06T21:57:01.768655Z","shell.execute_reply":"2021-12-06T21:57:01.774845Z"},"trusted":true},"execution_count":10,"outputs":[{"execution_count":10,"output_type":"execute_result","data":{"text/plain":"{'max_depth': 100,\n 'max_features': 'auto',\n 'max_leaf_nodes': 100,\n 'min_samples_leaf': 1,\n 'min_samples_split': 100,\n 'min_weight_fraction_leaf': 0,\n 'random_state': 0}"},"metadata":{}}]},{"cell_type":"code","source":"# Use with hypertuned parameters\ndt = DecisionTreeClassifier(max_depth=100,max_features='auto',max_leaf_nodes=100,min_samples_leaf=1,min_samples_split=100,min_weight_fraction_leaf=0, random_state=0)\ndt.fit(X_train, y_train)","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:57:01.776918Z","iopub.execute_input":"2021-12-06T21:57:01.777155Z","iopub.status.idle":"2021-12-06T21:57:01.965556Z","shell.execute_reply.started":"2021-12-06T21:57:01.777125Z","shell.execute_reply":"2021-12-06T21:57:01.964676Z"},"trusted":true},"execution_count":11,"outputs":[{"execution_count":11,"output_type":"execute_result","data":{"text/plain":"DecisionTreeClassifier(max_depth=100, max_features='auto', max_leaf_nodes=100,\n                       min_samples_split=100, min_weight_fraction_leaf=0,\n                       random_state=0)"},"metadata":{}}]},{"cell_type":"code","source":"# make prediction\ny_predict = dt.predict(X_test)\ndt.score(X_train, y_train) * 100","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:57:01.966850Z","iopub.execute_input":"2021-12-06T21:57:01.967085Z","iopub.status.idle":"2021-12-06T21:57:02.031760Z","shell.execute_reply.started":"2021-12-06T21:57:01.967054Z","shell.execute_reply":"2021-12-06T21:57:02.030787Z"},"trusted":true},"execution_count":12,"outputs":[{"execution_count":12,"output_type":"execute_result","data":{"text/plain":"69.78125"},"metadata":{}}]},{"cell_type":"code","source":"# test prediction\nprint(y_predict[0:20])\nprint(y_test[0:20])\nprint(metrics.accuracy_score(y_test, y_predict) * 100)","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:57:02.035214Z","iopub.execute_input":"2021-12-06T21:57:02.035451Z","iopub.status.idle":"2021-12-06T21:57:02.047619Z","shell.execute_reply.started":"2021-12-06T21:57:02.035421Z","shell.execute_reply":"2021-12-06T21:57:02.046750Z"},"trusted":true},"execution_count":13,"outputs":[{"name":"stdout","text":"[1 0 1 7 0 0 7 5 3 0 7 3 1 8 3 1 3 5 7 8]\n[1, 0, 1, 4, 0, 0, 7, 3, 5, 3, 8, 9, 1, 3, 3, 1, 2, 0, 7, 5]\n68.05\n","output_type":"stream"}]},{"cell_type":"code","source":"dt_score = cross_val_score(dt, X_train, y_train, scoring=\"neg_mean_squared_error\",cv=1000)\ndt_score_rmse = np.sqrt(-dt_score)\ndt_score_rmse.mean()","metadata":{"execution":{"iopub.status.busy":"2021-12-06T21:57:02.049016Z","iopub.execute_input":"2021-12-06T21:57:02.049704Z","iopub.status.idle":"2021-12-06T22:02:25.374461Z","shell.execute_reply.started":"2021-12-06T21:57:02.049670Z","shell.execute_reply":"2021-12-06T22:02:25.373356Z"},"trusted":true},"execution_count":14,"outputs":[{"execution_count":14,"output_type":"execute_result","data":{"text/plain":"2.1953578552890813"},"metadata":{}}]},{"cell_type":"code","source":"dt_score_rmse.std()","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:02:25.376221Z","iopub.execute_input":"2021-12-06T22:02:25.376473Z","iopub.status.idle":"2021-12-06T22:02:25.383303Z","shell.execute_reply.started":"2021-12-06T22:02:25.376440Z","shell.execute_reply":"2021-12-06T22:02:25.382635Z"},"trusted":true},"execution_count":15,"outputs":[{"execution_count":15,"output_type":"execute_result","data":{"text/plain":"0.45414495177258524"},"metadata":{}}]},{"cell_type":"code","source":"cm= metrics.confusion_matrix(y_train,dt.predict(X_train))\ncm","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:02:25.384483Z","iopub.execute_input":"2021-12-06T22:02:25.385277Z","iopub.status.idle":"2021-12-06T22:02:25.489458Z","shell.execute_reply.started":"2021-12-06T22:02:25.385233Z","shell.execute_reply":"2021-12-06T22:02:25.488462Z"},"trusted":true},"execution_count":16,"outputs":[{"execution_count":16,"output_type":"execute_result","data":{"text/plain":"array([[2719,    6,   82,   60,    1,  114,   26,   58,   68,    7],\n       [   1, 3169,  133,   34,   59,   47,   19,   90,   31,    6],\n       [ 126,   97, 2081,  157,   67,   83,  104,  101,  293,   23],\n       [  86,   53,  208, 2078,   25,  447,   65,  142,  172,   66],\n       [  21,   33,   39,   29, 2019,   94,  115,  160,  126,  469],\n       [ 160,   42,  120,  625,   66, 1439,  113,  104,  150,   70],\n       [ 118,   51,  270,   68,   43,  184, 2110,   87,  194,    9],\n       [  36,   38,   48,   23,   88,   91,   25, 2762,   29,  222],\n       [  91,  137,  139,  309,   99,  209,  144,  123, 1717,  145],\n       [  29,   19,   41,  128,   97,  142,   34,  314,  153, 2236]])"},"metadata":{}}]},{"cell_type":"code","source":"plt.figure(figsize=(14,12))\nplt.imshow(cm,cmap=plt.cm.Blues)\nplt.title(\"Decision Tree Confusion Matrix for MNIST Data\")\nplt.xticks(np.arange(10))\nplt.yticks(np.arange(10))\nplt.ylabel('Actual Label')\nplt.xlabel('Predicted Label')\nplt.colorbar()\nwidth,height = cm.shape\nfor x in range(width):\n    for y in range(height):\n        plt.annotate(str(cm[x][y]),xy=(y,x),horizontalalignment='center',verticalalignment='center')\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:02:25.491091Z","iopub.execute_input":"2021-12-06T22:02:25.491343Z","iopub.status.idle":"2021-12-06T22:02:26.556528Z","shell.execute_reply.started":"2021-12-06T22:02:25.491307Z","shell.execute_reply":"2021-12-06T22:02:26.555881Z"},"trusted":true},"execution_count":17,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1008x864 with 2 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"y_predict = dt.predict(X_test)\n\n# Actual and predicted classes\nlst_actual_class = y_test\nlst_predicted_class = y_predict\n\n# Class = digit labels 0, 1, 2, 3, 4, 5, 6, 7, 8, 9\nlst_classes = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] \n\n# Compute multi-class confusion matrix\narr_out_matrix = multilabel_confusion_matrix(lst_actual_class, lst_predicted_class, labels=lst_classes)\n\n# Temp store results\nstore_sens = [];\nstore_spec = [];\nstore_acc = [];\nstore_bal_acc = [];\nstore_prec = [];\nstore_fscore = [];\nstore_mcc = [];\n\n# Loop for each target label\nfor no_class in range(len(lst_classes)):\n    arr_data = arr_out_matrix[no_class];\n    print(\"Predicted Performance of Digit Label/Class: {0}\".format(no_class));\n\n    tp = arr_data[1][1] # true positive\n    fp = arr_data[0][1] # false positive\n    tn = arr_data[0][0] # true negative\n    fn = arr_data[1][0] # false negative\n\n    sensitivity = round(tp/(tp+fn), 3);\n    specificity = round(tn/(tn+fp), 3);\n    accuracy = round((tp+tn)/(tp+fp+tn+fn), 3);\n    balanced_accuracy = round((sensitivity+specificity)/2, 3);\n    precision = round(tp/(tp + fp), 3)\n    fscore = round((2 * ((precision * sensitivity) / (precision + sensitivity))), 3)\n    mcc = round((((tp * tn)-(fp * fn))/ math.sqrt((tp + fp)*(tp + fn)*(tn + fp)*(tn + fn))),3)\n    \n    store_sens.append(sensitivity);\n    store_spec.append(specificity);\n    store_acc.append(accuracy);\n    store_bal_acc.append(balanced_accuracy);\n    store_prec.append(precision);\n    store_fscore.append(fscore);\n    store_mcc.append(mcc);\n    \n    print(\"TP={0}, FP={1}, TN={2}, FN={3}\".format(tp, fp, tn, fn));\n    print(\"Sensitivity: {0}\".format(sensitivity));\n    print(\"Specificity: {0}\".format(specificity));\n    print(\"Accuracy: {0}\".format(accuracy));\n    print(\"Balanced Accuracy: {0}\".format(balanced_accuracy));\n    print(\"Precision: {0}\".format(precision));\n    print(\"F1-Score: {0}\".format(fscore))\n    print(\"MCC: {0}\\n\".format(mcc))","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:02:26.557828Z","iopub.execute_input":"2021-12-06T22:02:26.558243Z","iopub.status.idle":"2021-12-06T22:02:26.612388Z","shell.execute_reply.started":"2021-12-06T22:02:26.558195Z","shell.execute_reply":"2021-12-06T22:02:26.611368Z"},"trusted":true},"execution_count":18,"outputs":[{"name":"stdout","text":"Predicted Performance of Digit Label/Class: 0\nTP=855, FP=223, TN=8786, FN=136\nSensitivity: 0.863\nSpecificity: 0.975\nAccuracy: 0.964\nBalanced Accuracy: 0.919\nPrecision: 0.793\nF1-Score: 0.827\nMCC: 0.807\n\nPredicted Performance of Digit Label/Class: 1\nTP=972, FP=148, TN=8757, FN=123\nSensitivity: 0.888\nSpecificity: 0.983\nAccuracy: 0.973\nBalanced Accuracy: 0.936\nPrecision: 0.868\nF1-Score: 0.878\nMCC: 0.862\n\nPredicted Performance of Digit Label/Class: 2\nTP=665, FP=365, TN=8590, FN=380\nSensitivity: 0.636\nSpecificity: 0.959\nAccuracy: 0.926\nBalanced Accuracy: 0.798\nPrecision: 0.646\nF1-Score: 0.641\nMCC: 0.599\n\nPredicted Performance of Digit Label/Class: 3\nTP=630, FP=495, TN=8496, FN=379\nSensitivity: 0.624\nSpecificity: 0.945\nAccuracy: 0.913\nBalanced Accuracy: 0.784\nPrecision: 0.56\nF1-Score: 0.59\nMCC: 0.543\n\nPredicted Performance of Digit Label/Class: 4\nTP=595, FP=189, TN=8844, FN=372\nSensitivity: 0.615\nSpecificity: 0.979\nAccuracy: 0.944\nBalanced Accuracy: 0.797\nPrecision: 0.759\nF1-Score: 0.679\nMCC: 0.654\n\nPredicted Performance of Digit Label/Class: 5\nTP=412, FP=430, TN=8664, FN=494\nSensitivity: 0.455\nSpecificity: 0.953\nAccuracy: 0.908\nBalanced Accuracy: 0.704\nPrecision: 0.489\nF1-Score: 0.471\nMCC: 0.421\n\nPredicted Performance of Digit Label/Class: 6\nTP=643, FP=230, TN=8767, FN=360\nSensitivity: 0.641\nSpecificity: 0.974\nAccuracy: 0.941\nBalanced Accuracy: 0.808\nPrecision: 0.737\nF1-Score: 0.686\nMCC: 0.655\n\nPredicted Performance of Digit Label/Class: 7\nTP=833, FP=384, TN=8577, FN=206\nSensitivity: 0.802\nSpecificity: 0.957\nAccuracy: 0.941\nBalanced Accuracy: 0.88\nPrecision: 0.684\nF1-Score: 0.738\nMCC: 0.708\n\nPredicted Performance of Digit Label/Class: 8\nTP=501, FP=405, TN=8645, FN=449\nSensitivity: 0.527\nSpecificity: 0.955\nAccuracy: 0.915\nBalanced Accuracy: 0.741\nPrecision: 0.553\nF1-Score: 0.54\nMCC: 0.493\n\nPredicted Performance of Digit Label/Class: 9\nTP=699, FP=326, TN=8679, FN=296\nSensitivity: 0.703\nSpecificity: 0.964\nAccuracy: 0.938\nBalanced Accuracy: 0.833\nPrecision: 0.682\nF1-Score: 0.692\nMCC: 0.658\n\n","output_type":"stream"}]},{"cell_type":"markdown","source":"We see that for this subset of data with 10,000 rows, using the Decision Tree Classifier, for the '9' label we achieve an accuracy of 93.8 and precision of 68.2.","metadata":{}},{"cell_type":"code","source":"# use random forest algorithm next to compare to. Training time will be longer\nparameters = {'max_features': ('auto', 'sqrt', 'log2'), 'max_depth':[1,100], 'min_samples_split':[2,100], 'min_samples_leaf': [1,100], 'min_weight_fraction_leaf': [0,0.5], 'random_state': [0,100], 'max_leaf_nodes':[0,100]}\nmodel = RandomForestClassifier()\nclf = GridSearchCV(model, parameters)\nclf.fit(X_train, y_train)\nclf.score(X_train, y_train) * 100","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:02:26.613730Z","iopub.execute_input":"2021-12-06T22:02:26.613979Z","iopub.status.idle":"2021-12-06T22:29:39.096954Z","shell.execute_reply.started":"2021-12-06T22:02:26.613947Z","shell.execute_reply":"2021-12-06T22:29:39.095380Z"},"trusted":true},"execution_count":19,"outputs":[{"execution_count":19,"output_type":"execute_result","data":{"text/plain":"92.146875"},"metadata":{}}]},{"cell_type":"code","source":"clf.best_params_","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:29:39.098406Z","iopub.execute_input":"2021-12-06T22:29:39.098710Z","iopub.status.idle":"2021-12-06T22:29:39.106670Z","shell.execute_reply.started":"2021-12-06T22:29:39.098676Z","shell.execute_reply":"2021-12-06T22:29:39.105639Z"},"trusted":true},"execution_count":20,"outputs":[{"execution_count":20,"output_type":"execute_result","data":{"text/plain":"{'max_depth': 100,\n 'max_features': 'auto',\n 'max_leaf_nodes': 100,\n 'min_samples_leaf': 1,\n 'min_samples_split': 100,\n 'min_weight_fraction_leaf': 0,\n 'random_state': 0}"},"metadata":{}}]},{"cell_type":"code","source":"# Use with hypertuned parameters\nrf = RandomForestClassifier(max_depth=100,max_features='auto',min_samples_leaf=1,min_samples_split=100,min_weight_fraction_leaf=0, random_state=0)\nrf.fit(X_train, y_train)","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:29:39.108113Z","iopub.execute_input":"2021-12-06T22:29:39.108384Z","iopub.status.idle":"2021-12-06T22:29:56.050687Z","shell.execute_reply.started":"2021-12-06T22:29:39.108352Z","shell.execute_reply":"2021-12-06T22:29:56.050044Z"},"trusted":true},"execution_count":21,"outputs":[{"execution_count":21,"output_type":"execute_result","data":{"text/plain":"RandomForestClassifier(max_depth=100, min_samples_split=100,\n                       min_weight_fraction_leaf=0, random_state=0)"},"metadata":{}}]},{"cell_type":"code","source":"# make prediction\ny_predict = rf.predict(X_test)\nrf.score(X_train, y_train) * 100","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:29:56.051778Z","iopub.execute_input":"2021-12-06T22:29:56.052205Z","iopub.status.idle":"2021-12-06T22:29:57.392028Z","shell.execute_reply.started":"2021-12-06T22:29:56.052171Z","shell.execute_reply":"2021-12-06T22:29:57.390560Z"},"trusted":true},"execution_count":22,"outputs":[{"execution_count":22,"output_type":"execute_result","data":{"text/plain":"95.971875"},"metadata":{}}]},{"cell_type":"code","source":"# test prediction\nprint(y_predict[0:20])\nprint(y_test[0:20])\nprint(metrics.accuracy_score(y_test, y_predict) * 100)","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:29:57.393807Z","iopub.execute_input":"2021-12-06T22:29:57.394092Z","iopub.status.idle":"2021-12-06T22:29:57.408824Z","shell.execute_reply.started":"2021-12-06T22:29:57.394058Z","shell.execute_reply":"2021-12-06T22:29:57.407929Z"},"trusted":true},"execution_count":23,"outputs":[{"name":"stdout","text":"[1 0 1 4 0 0 7 3 5 3 8 9 1 3 3 1 8 0 7 8]\n[1, 0, 1, 4, 0, 0, 7, 3, 5, 3, 8, 9, 1, 3, 3, 1, 2, 0, 7, 5]\n93.89\n","output_type":"stream"}]},{"cell_type":"code","source":"cm= metrics.confusion_matrix(y_train,rf.predict(X_train))\ncm","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:29:57.410870Z","iopub.execute_input":"2021-12-06T22:29:57.411932Z","iopub.status.idle":"2021-12-06T22:29:58.545659Z","shell.execute_reply.started":"2021-12-06T22:29:57.411695Z","shell.execute_reply":"2021-12-06T22:29:58.544605Z"},"trusted":true},"execution_count":24,"outputs":[{"execution_count":24,"output_type":"execute_result","data":{"text/plain":"array([[3079,    0,    5,    0,    3,    7,   13,    0,   32,    2],\n       [   0, 3520,   19,   11,    6,    8,    4,    5,   13,    3],\n       [  11,    1, 3004,   18,   30,    3,   18,   20,   21,    6],\n       [  11,    7,   55, 3131,    7,   29,    5,   31,   44,   22],\n       [   6,    4,    8,    0, 3000,    1,   15,    7,   17,   47],\n       [  20,    5,    3,   43,    5, 2753,   20,    1,   21,   18],\n       [  19,    6,    3,    0,    5,   25, 3061,    0,   15,    0],\n       [   3,   11,   44,    2,   17,    0,    0, 3212,   12,   61],\n       [   7,   17,   15,   38,   13,   19,    8,    3, 2959,   34],\n       [  16,    8,    9,   54,   39,    6,    2,   35,   32, 2992]])"},"metadata":{}}]},{"cell_type":"code","source":"plt.figure(figsize=(14,12))\nplt.imshow(cm,cmap='rainbow_r')\nplt.title(\"Random Forest Confusion Matrix for MNIST Data\")\nplt.xticks(np.arange(10))\nplt.yticks(np.arange(10))\nplt.ylabel('Actual Label')\nplt.xlabel('Predicted Label')\nplt.colorbar()\nwidth,height = cm.shape\nfor x in range(width):\n    for y in range(height):\n        plt.annotate(str(cm[x][y]),xy=(y,x),horizontalalignment='center',verticalalignment='center')\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:29:58.547089Z","iopub.execute_input":"2021-12-06T22:29:58.547358Z","iopub.status.idle":"2021-12-06T22:29:59.633760Z","shell.execute_reply.started":"2021-12-06T22:29:58.547325Z","shell.execute_reply":"2021-12-06T22:29:59.632650Z"},"trusted":true},"execution_count":25,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1008x864 with 2 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"Random Forest performed much better than Decision Tree Classifier; this is because it is not relying on feature importance from one decision tree, but utilizing multiple ones generally. Random Forest should be used especially due to the size of the dataset.","metadata":{}},{"cell_type":"code","source":"# Actual and predicted classes\nlst_actual_class = y_test\nlst_predicted_class = y_predict\n\n# Class = digit labels 0, 1, 2, 3, 4, 5, 6, 7, 8, 9\nlst_classes = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] \n\n# Compute multi-class confusion matrix\narr_out_matrix = multilabel_confusion_matrix(lst_actual_class, lst_predicted_class, labels=lst_classes)\n\n# Temp store results\nstore_sens = [];\nstore_spec = [];\nstore_acc = [];\nstore_bal_acc = [];\nstore_prec = [];\nstore_fscore = [];\nstore_mcc = [];\n\n# Loop for each target label\nfor no_class in range(len(lst_classes)):\n    arr_data = arr_out_matrix[no_class];\n    print(\"Predicted Performance of Digit Label/Class: {0}\".format(no_class));\n\n    tp = arr_data[1][1] # true positive\n    fp = arr_data[0][1] # false positive\n    tn = arr_data[0][0] # true negative\n    fn = arr_data[1][0] # false negative\n\n    sensitivity = round(tp/(tp+fn), 3);\n    specificity = round(tn/(tn+fp), 3);\n    accuracy = round((tp+tn)/(tp+fp+tn+fn), 3);\n    balanced_accuracy = round((sensitivity+specificity)/2, 3);\n    precision = round(tp/(tp + fp), 3)\n    fscore = round((2 * ((precision * sensitivity) / (precision + sensitivity))), 3)\n    mcc = round((((tp * tn)-(fp * fn))/ math.sqrt((tp + fp)*(tp + fn)*(tn + fp)*(tn + fn))),3)\n    \n    store_sens.append(sensitivity);\n    store_spec.append(specificity);\n    store_acc.append(accuracy);\n    store_bal_acc.append(balanced_accuracy);\n    store_prec.append(precision);\n    store_fscore.append(fscore);\n    store_mcc.append(mcc);\n    \n    print(\"TP={0}, FP={1}, TN={2}, FN={3}\".format(tp, fp, tn, fn));\n    print(\"Sensitivity: {0}\".format(sensitivity));\n    print(\"Specificity: {0}\".format(specificity));\n    print(\"Accuracy: {0}\".format(accuracy));\n    print(\"Balanced Accuracy: {0}\".format(balanced_accuracy));\n    print(\"Precision: {0}\".format(precision));\n    print(\"F1-Score: {0}\".format(fscore))\n    print(\"MCC: {0}\\n\".format(mcc))","metadata":{"execution":{"iopub.status.busy":"2021-12-06T22:29:59.635671Z","iopub.execute_input":"2021-12-06T22:29:59.636312Z","iopub.status.idle":"2021-12-06T22:29:59.676605Z","shell.execute_reply.started":"2021-12-06T22:29:59.636273Z","shell.execute_reply":"2021-12-06T22:29:59.675693Z"},"trusted":true},"execution_count":26,"outputs":[{"name":"stdout","text":"Predicted Performance of Digit Label/Class: 0\nTP=965, FP=47, TN=8962, FN=26\nSensitivity: 0.974\nSpecificity: 0.995\nAccuracy: 0.993\nBalanced Accuracy: 0.984\nPrecision: 0.954\nF1-Score: 0.964\nMCC: 0.96\n\nPredicted Performance of Digit Label/Class: 1\nTP=1064, FP=28, TN=8877, FN=31\nSensitivity: 0.972\nSpecificity: 0.997\nAccuracy: 0.994\nBalanced Accuracy: 0.984\nPrecision: 0.974\nF1-Score: 0.973\nMCC: 0.97\n\nPredicted Performance of Digit Label/Class: 2\nTP=983, FP=73, TN=8882, FN=62\nSensitivity: 0.941\nSpecificity: 0.992\nAccuracy: 0.986\nBalanced Accuracy: 0.966\nPrecision: 0.931\nF1-Score: 0.936\nMCC: 0.928\n\nPredicted Performance of Digit Label/Class: 3\nTP=925, FP=73, TN=8918, FN=84\nSensitivity: 0.917\nSpecificity: 0.992\nAccuracy: 0.984\nBalanced Accuracy: 0.954\nPrecision: 0.927\nF1-Score: 0.922\nMCC: 0.913\n\nPredicted Performance of Digit Label/Class: 4\nTP=918, FP=63, TN=8970, FN=49\nSensitivity: 0.949\nSpecificity: 0.993\nAccuracy: 0.989\nBalanced Accuracy: 0.971\nPrecision: 0.936\nF1-Score: 0.942\nMCC: 0.936\n\nPredicted Performance of Digit Label/Class: 5\nTP=840, FP=62, TN=9032, FN=66\nSensitivity: 0.927\nSpecificity: 0.993\nAccuracy: 0.987\nBalanced Accuracy: 0.96\nPrecision: 0.931\nF1-Score: 0.929\nMCC: 0.922\n\nPredicted Performance of Digit Label/Class: 6\nTP=966, FP=48, TN=8949, FN=37\nSensitivity: 0.963\nSpecificity: 0.995\nAccuracy: 0.992\nBalanced Accuracy: 0.979\nPrecision: 0.953\nF1-Score: 0.958\nMCC: 0.953\n\nPredicted Performance of Digit Label/Class: 7\nTP=954, FP=35, TN=8926, FN=85\nSensitivity: 0.918\nSpecificity: 0.996\nAccuracy: 0.988\nBalanced Accuracy: 0.957\nPrecision: 0.965\nF1-Score: 0.941\nMCC: 0.935\n\nPredicted Performance of Digit Label/Class: 8\nTP=871, FP=84, TN=8966, FN=79\nSensitivity: 0.917\nSpecificity: 0.991\nAccuracy: 0.984\nBalanced Accuracy: 0.954\nPrecision: 0.912\nF1-Score: 0.914\nMCC: 0.905\n\nPredicted Performance of Digit Label/Class: 9\nTP=903, FP=98, TN=8907, FN=92\nSensitivity: 0.908\nSpecificity: 0.989\nAccuracy: 0.981\nBalanced Accuracy: 0.948\nPrecision: 0.902\nF1-Score: 0.905\nMCC: 0.894\n\n","output_type":"stream"}]},{"cell_type":"markdown","source":"We see that for this subset of data with 10,000 rows, using the Random Forest Classifier, for the '9' label we achieve an accuracy of 98.1 and precision of 90.2. Compared to Decision Tree, both score higher, especially the precision.","metadata":{}},{"cell_type":"markdown","source":"Decision Tree performed best on the following labels in order respectively: \n1, 7, 0, 9, 6, 2, 3, 4, 8, 5\n\nRandom Forest performed best on the following labels in order respectively: \n1, 7, 3, 0, 6, 2, 4, 9, 8, 5\n\nBoth classifiers did best on 1 & 7, and worst on 8 & 5. \n\nThe label '9' was higher ranked in performance for Decision Tree, but Random Forest still had better performance with precision (TP/predicted yes) of 90.2 vs. 68.2 for Decision Tree.","metadata":{}},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}