Overview The statistics dashboard provides instant insights into the filtering results, including total records, filtered tasks, occurrence rates, and per-keyword breakdown.
Core Function Statistics generation is handled by gerar_estatisticas() in app.py:55:
def gerar_estatisticas ( df_original , df_filtrado , palavras_chave ): """Gera estatísticas simples dos dados""" total = len (df_original) filtrados = len (df_filtrado) stats = { 'total' : total, 'filtrados' : filtrados, 'percentual' : round ((filtrados / total) * 100 , 1 ) if total > 0 else 0 , 'por_palavra' : {} } for palavra in palavras_chave: if not df_filtrado.empty: count = int (df_filtrado[ 'Relato' ].str.contains( palavra, case = False , na = False ).sum()) if count > 0 : stats[ 'por_palavra' ][palavra] = count return stats
Metrics Provided
Total Records Count of all rows in the original file (before filtering)
Filtered Tasks Number of tasks matching at least one keyword
Occurrence Rate Percentage of filtered tasks relative to total records
Per-Keyword Breakdown Individual match count for each keyword used
Statistics Structure The function returns a dictionary with this structure:
{ 'total' : 1523 , # Total rows in original file 'filtrados' : 147 , # Rows matching keywords 'percentual' : 9.7 , # Percentage (1 decimal place) 'por_palavra' : { # Per-keyword counts 'quebrado' : 45 , 'solicitar peça' : 32 , 'trocar cabo' : 28 , 'danificado' : 18 , 'instalar' : 24 } }
Calculation Details
View Percentage Calculation Logic
percentual = round ((filtrados / total) * 100 , 1 ) if total > 0 else 0
Divides filtered count by total count
Multiplies by 100 to get percentage
Rounds to 1 decimal place using round()
Returns 0 if total is 0 (prevents division by zero)
Per-Keyword Counting Each keyword is counted independently using pandas string matching:
for palavra in palavras_chave: if not df_filtrado.empty: count = int (df_filtrado[ 'Relato' ].str.contains( palavra, case = False , na = False ).sum()) if count > 0 : stats[ 'por_palavra' ][palavra] = count
Keywords with zero matches are excluded from the por_palavra dictionary to keep the output clean.
Important Behaviors Overlapping Matches : A single task can match multiple keywords. The sum of per-keyword counts may exceed the total filtered count.Example: A task with “solicitar peça quebrada” matches both “solicitar peça” and “quebrado”.
Empty DataFrame Handling The function safely handles empty results:
if not df_filtrado.empty: # Only calculate per-keyword stats if there are results
This prevents errors when no tasks match any keywords.
Usage in Application Statistics are generated during file upload and stored in the Flask session:
# From app.py:141 stats = gerar_estatisticas(df_original, resultado_final, palavras_chave) session[ 'last_stats' ] = stats
Display in Results Page The stats dictionary is passed to the resultado.html template:
return render_template( 'resultado.html' , table = tabela_html, has_results = not resultado_final.empty, stats = stats, palavras_utilizadas = palavras_chave)
Statistics in Exports Statistics are exported to a dedicated “Estatísticas” sheet: stats_df = pd.DataFrame([ [ 'Total de Registros' , stats.get( 'total' , 'N/A' )], [ 'Tarefas Encontradas' , stats.get( 'filtrados' , 'N/A' )], [ 'Taxa de Ocorrência (%)' , stats.get( 'percentual' , 'N/A' )], [ 'Data de Geração' , datetime.now().strftime( ' %d /%m/%Y %H:%M' )] ], columns = [ 'Métrica' , 'Valor' ]) stats_df.to_excel(writer, index = False , sheet_name = 'Estatísticas' )
Statistics appear in a visual dashboard at the top of the PDF: < div class = "stats" > < div class = "stat-box" > < div class = "stat-number" > {stats.get('total', 'N/A')} </ div > < div > Total de Registros </ div > </ div > < div class = "stat-box" > < div class = "stat-number" > {stats.get('filtrados', 'N/A')} </ div > < div > Tarefas Encontradas </ div > </ div > < div class = "stat-box" > < div class = "stat-number" > {stats.get('percentual', 'N/A')}% </ div > < div > Taxa de Ocorrência </ div > </ div > </ div >
Data Types Field Type Example total int 1523 filtrados int 147 percentual float 9.7 por_palavra dict {"quebrado": 45}
The per_palavra counts are explicitly converted to int using int() to ensure consistent JSON serialization when storing in Flask sessions.
Statistics generation is fast even for large datasets:
Uses vectorized pandas operations
Runs in O(n × k) where n = filtered rows, k = number of keywords
Typically completes in less than 100ms for files with 10,000 rows