Spaces:
Sleeping
Sleeping
Daniel Varga
commited on
Commit
•
8ec3059
1
Parent(s):
7c98739
can add predictor backends
Browse files- demo_prophet.py +26 -14
demo_prophet.py
CHANGED
@@ -4,6 +4,7 @@ import matplotlib.pyplot as plt
|
|
4 |
from prophet import Prophet
|
5 |
import holidays
|
6 |
import logging
|
|
|
7 |
|
8 |
|
9 |
# kW
|
@@ -14,12 +15,8 @@ hungarian_holidays = holidays.Hungary(years=range(2019, 2031))
|
|
14 |
HOLIDAY_DF = pd.DataFrame(list(hungarian_holidays.items()), columns=['ds', 'holiday'])
|
15 |
|
16 |
|
17 |
-
def prediction_task(df, split_date, forecast_horizon):
|
18 |
-
# Split the data into training (past) and evaluation (future) sets
|
19 |
-
train_data = df[df['ds'] <= split_date]
|
20 |
-
eval_data = df[df['ds'] > split_date]
|
21 |
-
|
22 |
|
|
|
23 |
# Initialize and train the Prophet model using the training data
|
24 |
model = Prophet(seasonality_mode='multiplicative', growth='flat',
|
25 |
yearly_seasonality=False, weekly_seasonality=True, daily_seasonality=True,
|
@@ -32,19 +29,33 @@ def prediction_task(df, split_date, forecast_horizon):
|
|
32 |
|
33 |
# Make predictions for the evaluation period
|
34 |
forecast = model.predict(future)
|
|
|
35 |
|
36 |
-
# Calculate evaluation metrics (e.g., MAE, MSE, RMSE) by comparing eval_predictions with eval_data['y']
|
37 |
-
|
38 |
-
# For example, you can calculate MAE as follows:
|
39 |
-
from sklearn.metrics import mean_absolute_error
|
40 |
-
|
41 |
-
eval_data = eval_data[eval_data['ds'] <= forecast['ds'].max()]
|
42 |
for key in ('yhat', 'yhat_lower', 'yhat_upper'):
|
43 |
forecast[key] = np.maximum(forecast[key], PREDICTION_LOWER_BOUND)
|
44 |
|
45 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
|
47 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
|
49 |
do_vis = False
|
50 |
if do_vis:
|
@@ -66,6 +77,7 @@ def prediction_task(df, split_date, forecast_horizon):
|
|
66 |
|
67 |
fig2 = model.plot_components(forecast)
|
68 |
plt.show()
|
|
|
69 |
return mae, eval_data['y'].mean()
|
70 |
|
71 |
|
@@ -100,7 +112,7 @@ weekly_date_range = pd.date_range(start=start_date, end=end_date, freq='8d')
|
|
100 |
maes = []
|
101 |
mean_values = []
|
102 |
for split_date in weekly_date_range:
|
103 |
-
mae, mean_value = prediction_task(df, split_date, forecast_horizon)
|
104 |
maes.append(mae)
|
105 |
mean_values.append(mean_value)
|
106 |
print(split_date, "Mean Absolute Error", mae, "MAE/true mean", mae / mean_value)
|
|
|
4 |
from prophet import Prophet
|
5 |
import holidays
|
6 |
import logging
|
7 |
+
from sklearn.metrics import mean_absolute_error
|
8 |
|
9 |
|
10 |
# kW
|
|
|
15 |
HOLIDAY_DF = pd.DataFrame(list(hungarian_holidays.items()), columns=['ds', 'holiday'])
|
16 |
|
17 |
|
|
|
|
|
|
|
|
|
|
|
18 |
|
19 |
+
def prophet_backend(train_data, forecast_horizon):
|
20 |
# Initialize and train the Prophet model using the training data
|
21 |
model = Prophet(seasonality_mode='multiplicative', growth='flat',
|
22 |
yearly_seasonality=False, weekly_seasonality=True, daily_seasonality=True,
|
|
|
29 |
|
30 |
# Make predictions for the evaluation period
|
31 |
forecast = model.predict(future)
|
32 |
+
assert len(forecast) == forecast_horizon
|
33 |
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
for key in ('yhat', 'yhat_lower', 'yhat_upper'):
|
35 |
forecast[key] = np.maximum(forecast[key], PREDICTION_LOWER_BOUND)
|
36 |
|
37 |
+
return forecast
|
38 |
+
|
39 |
+
|
40 |
+
def sklearn_backend(train_data, forecast_horizon):
|
41 |
+
dc = train_data[['y']]
|
42 |
+
# inserting new column with yesterday's consumption values
|
43 |
+
for i in range(1, 4 * 24 + 1):
|
44 |
+
dc.loc[:, 'd%02d' % i] = dc.loc[:,'y'].shift(4 * 24 + i) # t-2days to t-1day
|
45 |
+
dc.loc[:, 'w%02d' % i] = dc.loc[:,'y'].shift(7 * 4 * 24 + i) # t-7days to t-8days
|
46 |
+
dc.info()
|
47 |
+
exit()
|
48 |
|
49 |
+
|
50 |
+
def prediction_task(backend, df, split_date, forecast_horizon):
|
51 |
+
# Split the data into training (past) and evaluation (future) sets
|
52 |
+
train_data = df[df['ds'] <= split_date]
|
53 |
+
eval_data = df[df['ds'] > split_date]
|
54 |
+
eval_data = eval_data.head(forecast_horizon)
|
55 |
+
|
56 |
+
forecast = backend(train_data, forecast_horizon)
|
57 |
+
|
58 |
+
mae = mean_absolute_error(eval_data['y'], forecast['yhat'])
|
59 |
|
60 |
do_vis = False
|
61 |
if do_vis:
|
|
|
77 |
|
78 |
fig2 = model.plot_components(forecast)
|
79 |
plt.show()
|
80 |
+
|
81 |
return mae, eval_data['y'].mean()
|
82 |
|
83 |
|
|
|
112 |
maes = []
|
113 |
mean_values = []
|
114 |
for split_date in weekly_date_range:
|
115 |
+
mae, mean_value = prediction_task(prophet_backend, df, split_date, forecast_horizon)
|
116 |
maes.append(mae)
|
117 |
mean_values.append(mean_value)
|
118 |
print(split_date, "Mean Absolute Error", mae, "MAE/true mean", mae / mean_value)
|