This notebook is part of a GitHub repository: https://github.com/pessini/moby-bikes
MIT Licensed
Author: Leandro Pessini
In [21]:
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np
import joblib
import pickle
# Preprocessing
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler, MinMaxScaler, OneHotEncoder, OrdinalEncoder
from sklearn.compose import ColumnTransformer
from sklearn.pipeline import Pipeline
# metrics
from sklearn import metrics
# Boost models
import xgboost as xgb
from sklearn.ensemble import RandomForestRegressor
import time
import warnings
warnings.simplefilter('ignore', FutureWarning)
Evaluation¶
In [2]:
df_test = pd.read_csv('../data/processed/df_test.csv')
df_test.head()
Out[2]:
In [3]:
df = df_test.copy()
X = df.drop(['count'], axis=1)
y = df.pop('count')
all_columns = list(X.columns)
X.shape
Out[3]:
In [4]:
import category_encoders as ce
def preprocessor(predictors: list) -> ColumnTransformer:
# Setting remainder='passthrough' will mean that all columns not specified in the list of “transformers”
# will be passed through without transformation, instead of being dropped
##################### Categorical variables #####################
all_cat_vars = ['timesofday','dayofweek','holiday','peak','hour','working_day','season','month']
cat_vars = [categorical_var for categorical_var in all_cat_vars if categorical_var in predictors]
# categorical variables
cat_pipe = Pipeline([
('encoder', OneHotEncoder(handle_unknown='ignore', sparse=False))
])
cat_encoder = 'cat', cat_pipe, cat_vars
##################### Numerical variables #####################
all_num_vars = ['rain', 'temp', 'rhum','wdsp','temp_r']
num_vars = [numerical_var for numerical_var in all_num_vars if numerical_var in predictors]
num_pipe = Pipeline([
('scaler', StandardScaler())
# ('scaler', MinMaxScaler())
])
num_enconder = 'num', num_pipe, num_vars
##################### Ordinal variables #####################
all_ord_vars = ['wind_speed_group','rainfall_intensity']
ord_vars = [ordinal_var for ordinal_var in all_ord_vars if ordinal_var in predictors]
ordinal_cols_mapping = []
if 'wind_speed_group' in predictors:
ordinal_cols_mapping.append(
{"col":"wind_speed_group",
"mapping": {
'Calm / Light Breeze': 0,
'Breeze': 1,
'Moderate Breeze': 2,
'Strong Breeze / Near Gale': 3,
'Gale / Storm': 4
}}
)
if 'rainfall_intensity' in predictors:
ordinal_cols_mapping.append(
{"col":"rainfall_intensity",
"mapping": {
'no rain': 0,
'drizzle': 1,
'light rain': 2,
'moderate rain': 3,
'heavy rain': 4
}}
)
# ordinal variables
ord_pipe = Pipeline([
('ordinal', ce.OrdinalEncoder(mapping=ordinal_cols_mapping))
])
ord_enconder = 'ordinal', ord_pipe, ord_vars
#################################################################################
orig_vars = [var for var in predictors if var not in cat_vars and var not in num_vars and var not in ord_vars]
orig_enconder = 'pass_vars', 'passthrough', orig_vars
# ['temp_bin','rhum_bin']
# ord_pipe = 'passthrough'
transformers_list = []
transformers_list.append(cat_encoder) if cat_vars else None
transformers_list.append(ord_enconder) if ord_vars else None
transformers_list.append(num_enconder) if num_vars else None
# transformers_list.append(orig_enconder) if orig_vars else None
return ColumnTransformer(transformers=transformers_list,
remainder='drop')
Loading models¶
In [5]:
pipeline_xgboost = joblib.load('../models/xgb_pipeline.pkl')
In [6]:
xgb_model = xgb.XGBRegressor()
xgb_model.load_model("../models/XGBoost.json")
In [7]:
xgb_pipe = pickle.load(open("../models/xgb_pipeline.pkl", "rb"))
In [8]:
xgb_pipe
Out[8]:
In [ ]:
# predictors = ['temp','rhum','dayofweek','timesofday','wdsp','rainfall_intensity', 'working_day', 'hour', 'season']
In [9]:
predicted = pd.Series(xgb_pipe.predict(X))
In [10]:
def round_up(x):
'''
Helper function to round away from zero
'''
from math import copysign
return int(x + copysign(0.5, x))
Normalized Root Mean Square Error (NRMSE)¶
$NRMSE = \frac{RSME}{y_{max} - y_{min}}$
In [11]:
# NRMSE (Normalized Root Mean Square Error)
def normalized_rsme(value, dataset) -> float:
return value / max(dataset['count']) - min(dataset['count'])
In [12]:
rsme_score = metrics.mean_squared_error(y, predicted, squared=False)
normalized_rsme_score = normalized_rsme(rsme_score, df_test)
mae_score = metrics.mean_absolute_error(y, predicted)
In [13]:
actual_predicted = {'Actual': y, 'Predicted': predicted}
new_df = pd.DataFrame(actual_predicted)
new_df["Period"] = pd.to_datetime(X.date) + pd.to_timedelta(X.hour, unit='h')
new_df.set_index('Period')
Out[13]:
In [14]:
new_df['Rounded'] = new_df['Predicted'].apply(round_up)
new_df['Rounded_up'] = new_df['Predicted'].apply(round)
In [15]:
new_df.head()
Out[15]:
In [16]:
new_df.describe()
Out[16]:
In [17]:
fig, ax = plt.subplots(figsize=(36, 16), dpi=150)
sns.lineplot(data=new_df, x='Period', y='Actual', ax=ax, ci=None)
sns.lineplot(data=new_df, x='Period', y='Predicted', ax=ax, ci=None)
ax.set(xlabel='Datetime', ylabel='', title='Actual vs Predicted')
plt.legend(title='', loc='upper left', labels=['Actual', 'Predicted'], fontsize=20, labelspacing=0.4)
plt.annotate('> 15% infected birds', xy=(12, 3), xytext=(10, 3),
fontsize=10,fontweight='bold',fontfamily='serif',ha='center', color='#004D44')
annotate_text = f'RSME = {round(rsme_score,5)}\nNRMSE = {round(normalized_rsme_score,5)}\nMAE = {round(mae_score,5)}'
plt.annotate(annotate_text, xy=(pd.to_datetime('2022-03-15'),16), xytext=(pd.to_datetime('2022-03-15'), 16),
fontsize=16,fontweight='bold',fontfamily='serif',ha='center')
plt.show()
Daily rentals "mean"¶
In [18]:
daily_meandf = new_df.groupby(new_df['Period'].dt.date).mean().reset_index()
fig, ax = plt.subplots(figsize=(16, 12))
sns.lineplot(data=daily_meandf, x='Period', y='Actual', ax=ax, ci=None)
sns.lineplot(data=daily_meandf, x='Period', y='Predicted', ax=ax, ci=None)
ax.set(xlabel='Datetime', ylabel='', title='Actual vs Predicted Daily Mean')
plt.legend(title='', loc='upper left', labels=['Actual', 'Predicted'], fontsize=20, labelspacing=0.4)
annotate_text = f'RSME = {round(rsme_score,5)}\nNRMSE = {round(normalized_rsme_score,5)}\nMAE = {round(mae_score,5)}'
plt.annotate(annotate_text, xy=(pd.to_datetime('2022-04-22'),2), xytext=(pd.to_datetime('2022-04-22'), 2),
fontsize=16,fontweight='bold',fontfamily='serif',ha='center')
plt.show()
Daily rentals "total"¶
In [19]:
daily_sumdf = new_df.groupby(new_df['Period'].dt.date).sum()
fig, ax = plt.subplots(figsize=(16, 12))
sns.lineplot(data=daily_sumdf, x='Period', y='Actual', ax=ax, ci=None)
sns.lineplot(data=daily_sumdf, x='Period', y='Predicted', ax=ax, ci=None)
ax.set(xlabel='Datetime', ylabel='', title='Actual vs Predicted Daily Total')
plt.legend(title='', loc='upper left', labels=['Actual', 'Predicted'], fontsize=20, labelspacing=0.4)
annotate_text = f'RSME = {round(rsme_score,5)}\nNRMSE = {round(normalized_rsme_score,5)}\nMAE = {round(mae_score,5)}'
plt.annotate(annotate_text, xy=(pd.to_datetime('2022-04-22'),50), xytext=(pd.to_datetime('2022-04-22'), 50),
fontsize=16,fontweight='bold',fontfamily='serif',ha='center')
plt.show()
In [20]:
%reload_ext watermark
%watermark -a "Leandro Pessini" -n -u -v -iv -w
GitHub repository
Author: Leandro Pessini