-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #2 from istat-methodology/dev
Refactored App
- Loading branch information
Showing
8 changed files
with
700 additions
and
487 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,170 @@ | ||
| import json | ||
| import polars as pl | ||
| import logging | ||
| from modules import params | ||
|
|
||
| logging.basicConfig( | ||
| level=logging.INFO, | ||
| format="%(asctime)s.%(msecs)03d %(levelname)s %(module)s - %(funcName)s: %(message)s", | ||
| datefmt="%Y-%m-%d %H:%M:%S", | ||
| ) | ||
|
|
||
| class BatchProcessing(): | ||
|
|
||
| def __init__(self): | ||
| self.dfs = {} | ||
| self.output = {} | ||
|
|
||
| def _load_data(self, path): | ||
| logging.info(f"[LOAD PARQUET] Loading Parquet file from \"{path}\"") | ||
| self.dfs['-'] = pl.read_parquet(path) | ||
| logging.info("[LOAD PARQUET] Parquet file loaded succesfully!") | ||
|
|
||
| def _classify_sentiment(self, score): | ||
| if score == None: | ||
| return None | ||
| if score <= params.SENTIMENT_THRESHOLDS['t1']: | ||
| sentiment = 'Strongly Negative' | ||
| return sentiment | ||
| elif score > params.SENTIMENT_THRESHOLDS['t1'] and score <= params.SENTIMENT_THRESHOLDS['t2']: | ||
| sentiment = 'Negative' | ||
| return sentiment | ||
| elif score > params.SENTIMENT_THRESHOLDS['t2'] and score <= params.SENTIMENT_THRESHOLDS['t3']: | ||
| sentiment = 'Neutral' | ||
| return sentiment | ||
| elif score > params.SENTIMENT_THRESHOLDS['t3'] and score <= params.SENTIMENT_THRESHOLDS['t4']: | ||
| sentiment = 'Positive' | ||
| return sentiment | ||
| else: | ||
| sentiment = 'Strongly Positive' | ||
| return sentiment | ||
|
|
||
| def _add_sentiment_class(self): | ||
| logging.info('[SENTIMENT_CLASS] Extracting sentiment classes...') | ||
| self.dfs['-'] = self.dfs['-'].with_columns( | ||
| pl.col('SENTIMENT_SCORE').map_elements( | ||
| self._classify_sentiment, return_dtype=pl.String | ||
| ).alias('SENTIMENT_CLASS') | ||
| ) | ||
| logging.info('[SENTIMENT CLASS] Sentiment classes extracted succesfully!') | ||
|
|
||
| def _get_subsets(self): | ||
| logging.info(f"[WORD LIST TO STRING] Converting word lists to strings...") | ||
| self.dfs['-'] = self.dfs['-'].with_columns( | ||
| pl.col("WORD_LIST").cast(pl.List(pl.String)).list.join(", ").alias("TEXT_STR") | ||
| ) | ||
| for group in params.SEMANTIC_GROUPS['GROUPS'].values(): | ||
| if group: | ||
| logging.info(f"[DATA FILTERING] Filtering series by group \"{group}\"...") | ||
| self.dfs[group] = self.dfs['-'].filter(pl.col('TEXT_STR').str.contains(group)) | ||
|
|
||
| def _aggregate_series(self): | ||
| self.aggregate_dfs = {} | ||
| for group, df in self.dfs.items(): | ||
| logging.info(f"[SERIES AGGREGATION] Aggregating series for group \"{group}\"...") | ||
| grouped_df = df.group_by('DATE', maintain_order=True).agg([ | ||
| pl.col('SENTIMENT_SCORE').mean(), | ||
| pl.col('SENTIMENT_CLASS').value_counts(), | ||
| pl.col('TEXT_STR').len().alias('TOTAL_COUNT') | ||
| ]) | ||
| sentiment_exploded = grouped_df.explode('SENTIMENT_CLASS').unnest('SENTIMENT_CLASS') | ||
| sentiment_pivot = sentiment_exploded.pivot( | ||
| values="count", | ||
| index="DATE", | ||
| on=["SENTIMENT_CLASS"], | ||
| aggregate_function="sum" | ||
| ) | ||
| sentiment_pivot = sentiment_pivot.with_columns( | ||
| [pl.col(col).cast(pl.Int64).alias(col) for col in df['SENTIMENT_CLASS'].unique()] | ||
| ) | ||
| self.aggregate_dfs[group] = sentiment_pivot.with_columns( | ||
| grouped_df["SENTIMENT_SCORE"], | ||
| (((sentiment_pivot["Positive"] + sentiment_pivot["Strongly Positive"]) - | ||
| (sentiment_pivot["Negative"] + sentiment_pivot["Strongly Negative"])) / | ||
| grouped_df["TOTAL_COUNT"]).alias('QUALITATIVE_SCORE'), | ||
| grouped_df["TOTAL_COUNT"].cast(pl.Int64) | ||
| ) | ||
|
|
||
| def _get_quarters(self): | ||
| logging.info("[EXTRACT QUARTERS] Extracting quarters...") | ||
| self.freq_dfs = {} | ||
| for group, df in self.dfs.items(): | ||
| self.freq_dfs[group] = {} | ||
| df = df.select(['DATE', 'SENTIMENT_CLASS', 'WORD_LIST']) | ||
| df = df.with_columns(pl.col("DATE").str.strptime(pl.Date, "%Y-%m-%d")) | ||
| self.freq_dfs[group] = df.with_columns(("Q" + pl.col("DATE").dt.quarter().cast(pl.Utf8) + pl.col("DATE").dt.year().cast(pl.Utf8)).alias("QUARTER")) | ||
|
|
||
| def _word_frequencies(self, n_top_words): | ||
| logging.info('[WORD FREQUENCY] Starting word frequency procedure.') | ||
| self.word_freqs = {} | ||
|
|
||
| for group, df in self.freq_dfs.items(): | ||
| self.word_freqs[group] = {} | ||
|
|
||
| if group == '-': | ||
| classes = list(df['SENTIMENT_CLASS'].unique()) + ['Total'] | ||
| else: | ||
| classes = ['Total'] | ||
|
|
||
| for quarter in df["QUARTER"].unique(): | ||
| self.word_freqs[group][quarter] = {} | ||
| quarter_df = df.filter(pl.col("QUARTER") == quarter) | ||
|
|
||
| for sent_class in classes: | ||
| logging.info(f'[WORD FREQUENCY] Processing "{sent_class} | {quarter}" for group "{group}"...') | ||
| self.word_freqs[group][quarter][sent_class] = {} | ||
|
|
||
| if sent_class != 'Total': | ||
| sent_df = quarter_df.filter(pl.col("SENTIMENT_CLASS") == sent_class) | ||
| else: | ||
| sent_df = quarter_df | ||
|
|
||
| # explode the word_list column | ||
| sent_df = sent_df.explode("WORD_LIST").rename({"WORD_LIST": "WORD"}) | ||
|
|
||
| # count the word frequencies | ||
| word_counts = sent_df.group_by(['QUARTER', 'WORD']).len().rename({'len': 'FREQUENCY'}) | ||
|
|
||
| # extract the top 30 words | ||
| top_words = ( | ||
| word_counts | ||
| .sort(["QUARTER", "FREQUENCY"], descending=[False, True]) | ||
| .group_by("QUARTER") | ||
| .head(n_top_words) | ||
| ) | ||
|
|
||
| # create a dictionary where "key" = word, and "value" = count | ||
| df_temp = top_words.select('WORD', 'FREQUENCY').to_dict(as_series=False) | ||
| self.word_freqs[group][quarter][sent_class] = dict(zip(df_temp['WORD'], df_temp['FREQUENCY'])) | ||
|
|
||
| def _export(self, output_dir: str): | ||
| logging.info(f'[EXPORT] Exporting data to \"{output_dir}\"...') | ||
| time_series = {} | ||
|
|
||
| for group, df in self.aggregate_dfs.items(): | ||
| time_series[group] = df.sort('DATE').to_dict(as_series=False) | ||
|
|
||
| self.output['time_series'] = time_series | ||
| self.output['word_frequencies'] = self.word_freqs | ||
|
|
||
| with open(output_dir + 'dashboard_data.json', 'w') as fp: | ||
| json.dump(self.output, fp) | ||
|
|
||
| def run(self, path_to_file: str, n_top_words: int = 30, output_dir: str = 'data/'): | ||
| self._load_data(path_to_file) | ||
| self._add_sentiment_class() | ||
| self._get_subsets() | ||
| self._aggregate_series() | ||
| self._get_quarters() | ||
| self._word_frequencies(n_top_words) | ||
| self._export(output_dir) | ||
| logging.info("[COMPLETED] Batch procedure completed succesfully!") | ||
|
|
||
|
|
||
| if __name__ == "__main__": | ||
| path_to_file = "data/full_polars_df.parquet" | ||
| output_dir = "data/" | ||
| n_top_words = 30 | ||
|
|
||
| processor = BatchProcessing() | ||
| processor.run(path_to_file, n_top_words=n_top_words, output_dir=output_dir) |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,22 +1,32 @@ | ||
| import streamlit as st | ||
| from modules import bits, tabs, params | ||
| import streamlit_nested_layout | ||
| from modules import bits, plots | ||
|
|
||
| bits.set_configs() | ||
| bits.set_session_states() | ||
| bits.set_header() | ||
| bits.set_session_states() | ||
| bits.load_data() | ||
| bits.global_filters() | ||
| bits.get_filtered_data() | ||
|
|
||
| overview_tab, frequency_tab = st.tabs(['Overview', 'Frequency Analysis']) | ||
|
|
||
| if st.session_state['data_ready']: | ||
| with overview_tab: | ||
| tabs.OverviewTab().add() | ||
| with overview_tab: | ||
| col1, col2 = st.columns(2) | ||
| with col1: | ||
| plots.VolumePlot().add() | ||
| with col2: | ||
| plots.QuantitativePlot().add() | ||
|
|
||
| with frequency_tab: | ||
| tabs.FrequencyTab().add() | ||
| with frequency_tab: | ||
| col1, col2 = st.columns(2) | ||
| with col1: | ||
| plots.QualitativePlot().add() | ||
| with col2: | ||
| plots.SentimentClassTS().add() | ||
|
|
||
| with st.sidebar: | ||
| tabs.Sidebar().add() | ||
| else: | ||
| st.warning(params.WARNINGS['no_data']) | ||
| col3, col4 = st.columns(2) | ||
| with col3: | ||
| plots.WordFrequencyPlot().add() | ||
| with col4: | ||
| plots.SentimentPie().add() |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.