Random forest Example

main.py

#!/usr/bin/python3

"""Random Forest Examples. 2020-09-17. Author: Yu JaeIL
This example follows https://towardsdatascience.com/random-forest-in-python-24d0893d51c0
These codes are written and test in python 3.7.7.

References,
decision tree:	https://www.youtube.com/watch?v=n0p0120Gxqk&t=5s
randorm forest:	https://www.youtube.com/watch?v=nZB37IBCiSA
One-Hot-Encode:	https://en.wikipedia.org/wiki/One-hot
Baseline:		https://developers.google.com/machine-learning/glossary#baseline
"""

"""Weather data for Seattle.
data fileds
	1. year:	2016 for all data points.
	2. month:	number for month of the year.
	3. day	: 	number for day of the year.
	4. week	:	day of the week as a character string.
	5. temp_2: 	max temperature 2 days prior.
	6. temp_1: 	max temperature 1 day prior.
	7. average:	historical average max temperature.
	8. actual:	max temperature measurement.
	9. friend:	friend's prediction, a random number between 20 below the average and 20 above the average.
"""

from pprint import pprint
import pandas as pd
import numpy as np

features = pd.read_csv('temps.csv')
print(features.head(5))
""" The output
   year  month  day  week  temp_2  temp_1  average  actual  forecast_noaa  forecast_acc  forecast_under  friend
0  2016      1    1   Fri      45      45     45.6      45             43            50              44      29
1  2016      1    2   Sat      44      45     45.7      44             41            50              44      61
2  2016      1    3   Sun      45      44     45.8      41             43            46              47      56
3  2016      1    4   Mon      44      41     45.9      40             44            48              46      53
4  2016      1    5  Tues      41      40     46.0      44             46            46              46      41
"""

print('The shape of our features is:', features.shape)
""" The output
The shape of our features is: (348, 12)
"""

# One-hot encode the data using pandas get_dummies
# I think pd.get_dummies function enum field make one-hots.
features = pd.get_dummies(features)

# Display the first 5 rows of the last 13 columns
print(features.iloc[:,5:].head(5))
"""The output
   average  actual  forecast_noaa  forecast_acc  forecast_under  friend  week_Fri  week_Mon  week_Sat  week_Sun  week_Thurs  week_Tues  week_Wed  
0     45.6      45             43            50              44      29         1         0         0         0           0          0         0  
1     45.7      44             41            50              44      61         0         0         1         0           0          0         0  
2     45.8      41             43            46              47      56         0         0         0         1           0          0         0  
3     45.9      40             44            48              46      53         0         1         0         0           0          0         0  
4     46.0      44             46            46              46      41         0         0         0         0           0          1         0  
"""

# Labels are the values we want to predict.(Answer)
labels = np.array(features['actual'])

# Remove the labels from the features
# axis 1 refers to the columns (it means axis 1 is column or 0 is row.)
features = features.drop('actual', axis = 1)

# Saving feature names for later use
feature_list = list(features.columns)

# Convert to numpy array
features = np.array(features)

# Using Skicit-learn to split data into training and testing sets
from sklearn.model_selection import train_test_split

# Split the data into training and testing sets
train_features, test_features, train_labels, test_labels = \
	train_test_split(features, labels, test_size = 0.25, random_state = 42)

print('%-25s' % 'Training Features Shape:', train_features.shape)
print('%-25s' % 'Training Labels Shape:', train_labels.shape)
print('%-25s' % 'Testing Features Shape:', test_features.shape)
print('%-25s' % 'Testing Labels Shape:', test_labels.shape)
"""Output
Training Features Shape:  (261, 17)
Training Labels Shape:    (261,)
Testing Features Shape:   (87, 17)
Testing Labels Shape:     (87,)
"""

# The baseline predications are the historical averages
baseline_preds = test_features[:, feature_list.index('average')]

# Baseline errors, and display average baseline error
baseline_errors = abs(baseline_preds - test_labels)

print('Average baseline error: ', round(np.mean(baseline_errors), 2))
"""The Output
Average baseline error:  5.06
"""

# Import the model we are using
from sklearn.ensemble import RandomForestRegressor

# Instantiate model with 1000 decision trees
rf = RandomForestRegressor(n_estimators = 1000, random_state = 42)

# Train the model on training data
rf.fit(train_features, train_labels)

# Use the forest's predict method on the test data
predictions = rf.predict(test_features)

# Calculate the absolute erros
erros = abs(predictions - test_labels)

# Print out the mean absolute error (mae)
print('Mean Absolute Error:', round(np.mean(erros), 2), 'degrees.')
"""The Output
Mean Absolute Error: 3.87 degrees.
"""

"""My Understanding How Randomforest can be appied to Machine Learning.
1. First make Data to Sclable Data. like these examples, weeks were changed
   7 features and have data 1 or 0 by one-hot-encoding.)
2. Make decision trees. and per condition on trees is logical operation having
   features as operand. (like, n-feature > weight)
3. And training weights. that is all. I think.
"""

temp.csv