# parity plot
else:
for j in Y.columns:
# training data
data_j = "Train"
data = predictions.loc[(predictions["Data"] == data_j)
& (predictions["Name"] == j)]
r2 = str(np.round(r2_score(data["Actual"], data["Predict"]) * 100,
1)) + "%"
parity_plot(predict=data["Predict"],
actual=data["Actual"],
title=j + " - " + data_j + " - R2: " + r2,
save=save_plot)
# testing data
data_j = "Test"
data = predictions.loc[(predictions["Data"] == data_j)
& (predictions["Name"] == j)]
r2 = str(np.round(r2_score(data["Actual"], data["Predict"]) * 100,
1)) + "%"
parity_plot(predict=data["Predict"],
actual=data["Actual"],
title=j + " - " + data_j + " - R2: " + r2,
save=save_plot)
# series plot
series_plot(predict=data["Predict"],
actual=data["Actual"],
title=j + " - " + data_j + " - Forecast: ",
save=save_plot)
predictions.append(yhat)
obs = test[t]
history.append(obs)
print('predicted=%f, expected=%f' % (yhat, obs))
# collect predictions and actual values
predictions = pd.DataFrame({
"Actual": test.flatten(),
"Predict": np.array(predictions).flatten()
})
# predictions.to_csv("holt predictions.csv", index=False)
# In[2]: Visualize the predictions
save_plot = False
# series plot
series_plot(predict=predictions["Predict"],
actual=predictions["Actual"],
title="Holt-Winter's One-Step Ahead Rolling Forecast - Test",
save=save_plot)
# parity plot
r2 = str(
np.round(r2_score(predictions["Actual"], predictions["Predict"]) * 100,
1)) + "%"
parity_plot(predict=predictions["Predict"],
actual=predictions["Actual"],
title="Holt-Winter's Parity Plot - Test - R2: " + r2,
save=save_plot)
pipeline.fit(X.iloc[train_idx, :], Y.iloc[train_idx, :])
# report forecast
y_pred = pred["t1"]
y_true = Y.iloc[train_idx[-1], :][0]
print('predicted=%f, expected=%f' % (y_pred, y_true))
# compute r2 for the predictions
r2 = []
for j in predictions.columns:
r2.append(r2_score(actual[j], predictions[j]))
# In[2]: Visualize the predictions
save_plot = False
# choose a time horizon
t = 1
# series plot
series_plot(predict=predictions.iloc[:, t - 1],
actual=actual.iloc[:, t - 1],
title="Lasso " + str(t) + "-Step Ahead Rolling Forecast - Test",
save=save_plot)
# parity plot
parity_plot(predict=predictions.iloc[:, t - 1],
actual=actual.iloc[:, t - 1],
title="Lasso Parity Plot - Test - R2: " + str(r2[t - 1]),
save=save_plot)
obs = test[t]
history.append(obs)
print('predicted=%f, expected=%f' % (yhat, obs))
# collect predictions and actual values
predictions = pd.DataFrame({
"Actual": test.flatten(),
"Predict": np.array(predictions).flatten()
})
predictions.to_csv("exp predictions.csv", index=False)
# In[2]: Visualize the predictions
save_plot = False
# series plot
series_plot(
predict=predictions["Predict"],
actual=predictions["Actual"],
title="Exponential Smoothing One-Step Ahead Rolling Forecast - Test",
save=save_plot)
# parity plot
r2 = str(
np.round(r2_score(predictions["Actual"], predictions["Predict"]) * 100,
1)) + "%"
parity_plot(predict=predictions["Predict"],
actual=predictions["Actual"],
title="Exponential Smoothing Parity Plot - Test - R2: " + r2,
save=save_plot)
