Update src/streamlit_app.py
Browse files- src/streamlit_app.py +23 -96
src/streamlit_app.py
CHANGED
|
@@ -164,10 +164,23 @@ try:
|
|
| 164 |
except Exception as e:
|
| 165 |
st.error(f"Could not decompose series: {e}")
|
| 166 |
|
| 167 |
-
# ACF / PACF Plots
|
| 168 |
st.subheader("🔗 Autocorrelation (ACF) & Partial Autocorrelation (PACF)")
|
| 169 |
|
| 170 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 171 |
|
| 172 |
col1, col2 = st.columns(2)
|
| 173 |
|
|
@@ -179,9 +192,13 @@ with col1:
|
|
| 179 |
|
| 180 |
with col2:
|
| 181 |
st.write("**PACF Plot**")
|
| 182 |
-
|
| 183 |
-
|
| 184 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 185 |
|
| 186 |
# === FORECASTING MODELS ===
|
| 187 |
st.header("🤖 Forecasting Models")
|
|
@@ -255,94 +272,4 @@ if forecast is not None:
|
|
| 255 |
# Metrics
|
| 256 |
mae = mean_absolute_error(test, forecast)
|
| 257 |
mse = mean_squared_error(test, forecast)
|
| 258 |
-
|
| 259 |
-
|
| 260 |
-
st.subheader("📈 Forecast Results")
|
| 261 |
-
col1, col2, col3 = st.columns(3)
|
| 262 |
-
col1.metric("MAE", f"{mae:.2f}")
|
| 263 |
-
col2.metric("MSE", f"{mse:.2f}")
|
| 264 |
-
col3.metric("RMSE", f"{rmse:.2f}")
|
| 265 |
-
|
| 266 |
-
# Plot forecast vs actual
|
| 267 |
-
fig = go.Figure()
|
| 268 |
-
fig.add_trace(go.Scatter(x=train.index, y=train, mode='lines', name='Training', line=dict(color='blue')))
|
| 269 |
-
fig.add_trace(go.Scatter(x=test.index, y=test, mode='lines', name='Actual', line=dict(color='green')))
|
| 270 |
-
fig.add_trace(go.Scatter(x=test.index, y=forecast, mode='lines+markers', name='Forecast', line=dict(color='red', dash='dash')))
|
| 271 |
-
fig.update_layout(
|
| 272 |
-
title=f"{model_choice} Forecast",
|
| 273 |
-
xaxis_title="Date",
|
| 274 |
-
yaxis_title=value_col,
|
| 275 |
-
legend=dict(x=0, y=1)
|
| 276 |
-
)
|
| 277 |
-
st.plotly_chart(fig, use_container_width=True)
|
| 278 |
-
|
| 279 |
-
# Allow forecasting into future
|
| 280 |
-
st.subheader("🔮 Forecast Future Periods")
|
| 281 |
-
future_periods = st.number_input("Number of future periods to forecast:", min_value=1, max_value=365, value=30, step=1)
|
| 282 |
-
|
| 283 |
-
if st.button("🚀 Generate Future Forecast"):
|
| 284 |
-
try:
|
| 285 |
-
if model_choice == "Holt-Winters Exponential Smoothing":
|
| 286 |
-
future_forecast = model.forecast(future_periods)
|
| 287 |
-
last_date = ts.index[-1]
|
| 288 |
-
if freq == 'D':
|
| 289 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(days=1), periods=future_periods, freq='D')
|
| 290 |
-
elif freq == 'W':
|
| 291 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(weeks=1), periods=future_periods, freq='W')
|
| 292 |
-
elif freq == 'M':
|
| 293 |
-
future_dates = pd.date_range(start=last_date + pd.DateOffset(months=1), periods=future_periods, freq='M')
|
| 294 |
-
else:
|
| 295 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(days=1), periods=future_periods, freq='D')
|
| 296 |
-
|
| 297 |
-
elif model_choice == "ARIMA":
|
| 298 |
-
future_forecast = model.forecast(future_periods)
|
| 299 |
-
last_date = ts.index[-1]
|
| 300 |
-
if freq == 'D':
|
| 301 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(days=1), periods=future_periods, freq='D')
|
| 302 |
-
elif freq == 'W':
|
| 303 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(weeks=1), periods=future_periods, freq='W')
|
| 304 |
-
elif freq == 'M':
|
| 305 |
-
future_dates = pd.date_range(start=last_date + pd.DateOffset(months=1), periods=future_periods, freq='M')
|
| 306 |
-
else:
|
| 307 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(days=1), periods=future_periods, freq='D')
|
| 308 |
-
|
| 309 |
-
elif model_choice == "Prophet":
|
| 310 |
-
last_date = ts.index[-1]
|
| 311 |
-
if freq == 'D':
|
| 312 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(days=1), periods=future_periods, freq='D')
|
| 313 |
-
elif freq == 'W':
|
| 314 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(weeks=1), periods=future_periods, freq='W')
|
| 315 |
-
elif freq == 'M':
|
| 316 |
-
future_dates = pd.date_range(start=last_date + pd.DateOffset(months=1), periods=future_periods, freq='M')
|
| 317 |
-
else:
|
| 318 |
-
future_dates = pd.date_range(start=last_date + pd.Timedelta(days=1), periods=future_periods, freq='D')
|
| 319 |
-
|
| 320 |
-
future_df = pd.DataFrame({'ds': future_dates})
|
| 321 |
-
forecast_df = model.predict(future_df)
|
| 322 |
-
future_forecast = forecast_df['yhat'].values
|
| 323 |
-
|
| 324 |
-
# Plot future forecast
|
| 325 |
-
fig_future = go.Figure()
|
| 326 |
-
fig_future.add_trace(go.Scatter(x=ts.index, y=ts.values, mode='lines', name='Historical', line=dict(color='blue')))
|
| 327 |
-
fig_future.add_trace(go.Scatter(x=future_dates, y=future_forecast, mode='lines+markers', name='Future Forecast', line=dict(color='red', dash='dash')))
|
| 328 |
-
fig_future.update_layout(
|
| 329 |
-
title="Future Forecast",
|
| 330 |
-
xaxis_title="Date",
|
| 331 |
-
yaxis_title=value_col
|
| 332 |
-
)
|
| 333 |
-
st.plotly_chart(fig_future, use_container_width=True)
|
| 334 |
-
|
| 335 |
-
# Show as table
|
| 336 |
-
forecast_df = pd.DataFrame({
|
| 337 |
-
'Date': future_dates,
|
| 338 |
-
'Forecast': future_forecast
|
| 339 |
-
})
|
| 340 |
-
with st.expander("📋 View Forecast Table"):
|
| 341 |
-
st.dataframe(forecast_df)
|
| 342 |
-
|
| 343 |
-
except Exception as e:
|
| 344 |
-
st.error(f"Could not generate future forecast: {e}")
|
| 345 |
-
|
| 346 |
-
# Footer
|
| 347 |
-
st.markdown("---")
|
| 348 |
-
st.caption(f"© {AUTHOR} | License {LICENSE} | Contact: {EMAIL}")
|
|
|
|
| 164 |
except Exception as e:
|
| 165 |
st.error(f"Could not decompose series: {e}")
|
| 166 |
|
| 167 |
+
# ACF / PACF Plots (CORREGIDO)
|
| 168 |
st.subheader("🔗 Autocorrelation (ACF) & Partial Autocorrelation (PACF)")
|
| 169 |
|
| 170 |
+
# Calculate safe max lags (must be < 50% of sample size for PACF)
|
| 171 |
+
n = len(ts.dropna())
|
| 172 |
+
safe_max_lag = max(1, int(n * 0.49)) # Must be strictly less than 50%
|
| 173 |
+
|
| 174 |
+
# Adjust slider dynamically
|
| 175 |
+
max_lags_default = min(40, safe_max_lag)
|
| 176 |
+
max_lags = st.slider(
|
| 177 |
+
"Max lags:",
|
| 178 |
+
min_value=1,
|
| 179 |
+
max_value=safe_max_lag,
|
| 180 |
+
value=max_lags_default,
|
| 181 |
+
step=1,
|
| 182 |
+
help=f"Max allowed lags: {safe_max_lag} (based on sample size: {n})"
|
| 183 |
+
)
|
| 184 |
|
| 185 |
col1, col2 = st.columns(2)
|
| 186 |
|
|
|
|
| 192 |
|
| 193 |
with col2:
|
| 194 |
st.write("**PACF Plot**")
|
| 195 |
+
try:
|
| 196 |
+
fig_pacf, ax_pacf = plt.subplots(figsize=(6, 4))
|
| 197 |
+
plot_pacf(ts.dropna(), lags=max_lags, ax=ax_pacf)
|
| 198 |
+
st.pyplot(fig_pacf)
|
| 199 |
+
except Exception as e:
|
| 200 |
+
st.error(f"Could not generate PACF plot: {e}")
|
| 201 |
+
st.write("Try reducing the number of lags.")
|
| 202 |
|
| 203 |
# === FORECASTING MODELS ===
|
| 204 |
st.header("🤖 Forecasting Models")
|
|
|
|
| 272 |
# Metrics
|
| 273 |
mae = mean_absolute_error(test, forecast)
|
| 274 |
mse = mean_squared_error(test, forecast)
|
| 275 |
+
rm
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|