samuelcochrane / cloudsml

Predicting the wheather with Python and Scikit

CloudsML

Predicting the wheather with Python and Scikit

bring in intial libraries we will need

 import numpy
 import pandas
 import sklearn

bring in our data

wheather = pandas.read_csv("C:/Users/Samue/Desktop/MyWebsite/CloudsML/seattleWeather_1948-2017Copy.csv", header = 0, sep = ",")

We can specify headers manually by adding names=["DATE", "PRCP", "TMAX", "TMIN", "RAIN"] to above

>>> print(wheather.head())

      DATE  PRCP  TMAX  TMIN  RAIN
0  19480101  0.47    51    42  True
1  19480102  0.59    45    36  True
2  19480103  0.42    45    35  True
3  19480104  0.31    45    34  True
4  19480105  0.17    45    32  True

>>> print(wheather.describe())

        DATE          PRCP          TMAX          TMIN
count  2.555100e+04  25548.000000  25551.000000  25551.000000
mean   1.982543e+07      0.106222     59.544206     44.514226
std    2.019306e+05      0.239031     12.772984      8.892836
min    1.948010e+07      0.000000      4.000000      0.000000
25%    1.965063e+07      0.000000     50.000000     38.000000
50%    1.982122e+07      0.000000     58.000000     45.000000
75%    2.000062e+07      0.100000     69.000000     52.000000
max    2.017121e+07      5.020000    103.000000     71.000000

>>> print(wheather.isnull().values.any())
True

We have a few days without rain data, such as "1998-06-02","NA",72,52,"NA". let's just remove them.

 wheather = wheather.dropna()

from sklearn.model_selection import train_test_split

We want to predict the RAIN value, so let's break out data into two sets. X, the independent data, and Y, the dependent data we want to predict.

X = wheather.drop(["PRCP", "RAIN"], axis=1) #all columns except PRCP and RAIN, as both give the answer away.
y = wheather["PRCP"]

Using the train_test_split function, we create the appropriate train/test data for our features ("X_train" and "X_test" respectively) and target data ("Y_train" and "Y_test"). We are specifying our test data to be 20% of the total data (80/20 split model, thanks Pareto). We are also providing a defined seed value (42) to be able to reproduce this split if we want to come back to it later.

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

>>> print(X_train.shape)
(20438, 3) #20438 training rows
>>> print(X_test.shape)
(5110, 3) #5110 testing rows

from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor

tree_model = DecisionTreeRegressor()
rf_model = RandomForestRegressor()

tree_model.fit(X_train, y_train)
rf_model.fit(X_train, y_train)

model evaluation

 from sklearn.metrics import mean_squared_error
 from sklearn.metrics import mean_absolute_error


 tree_mse = mean_squared_error(y_train, tree_model.predict(X_train))
 tree_mae = mean_absolute_error(y_train, tree_model.predict(X_train))
 rf_mse = mean_squared_error(y_train, rf_model.predict(X_train))
 rf_mae = mean_absolute_error(y_train, rf_model.predict(X_train))

 from math import sqrt

 print("Decision Tree training mse = ",tree_mse," & mae = ",tree_mae," & rmse = ", sqrt(tree_mse))
 print("Random Forest training mse = ",rf_mse," & mae = ",rf_mae," & rmse = ", sqrt(rf_mse))


 tree_test_mse = mean_squared_error(y_test, tree_model.predict(X_test))
 tree_test_mae = mean_absolute_error(y_test, tree_model.predict(X_test))
 rf_test_mse = mean_squared_error(y_test, rf_model.predict(X_test))
 rf_test_mae = mean_absolute_error(y_test, rf_model.predict(X_test))

 print("Decision Tree test mse = ",tree_test_mse," & mae = ",tree_test_mae," & rmse = ", sqrt(tree_test_mse))
 print("Random Forest test mse = ",rf_test_mse," & mae = ",rf_test_mae," & rmse = ", sqrt(rf_test_mse))

#Random forrest does better than decision tree, with an avg error of +-0.24 inches instead of +-0.29 inches of precipitation. Still, both aren't great. #Both are also doing much much better on the training data vs the test data, +-0.03 error vs 0.29 error is a big difference in inches.

 def display_scores(scores):
     print("Scores:", scores)
     print("Mean:", scores.mean())
     print("Standard deviation:", scores.std())
     print("\n")


 from sklearn.model_selection import cross_val_score

 scores = cross_val_score(tree_model, X_train, y_train, scoring="neg_mean_squared_error", cv=10)
 tree_rmse_scores = numpy.sqrt(-scores)

 scores = cross_val_score(rf_model, X_train, y_train, scoring="neg_mean_squared_error", cv=10)
 rf_rmse_scores = numpy.sqrt(-scores)

Let's see how our decision tree model preformed vs our random forest

>>> display_scores(tree_rmse_scores)
Scores: [0.32785747 0.3120757  0.31520675 0.3118263  0.28814834 0.31632178
 0.31382765 0.32981789 0.30310489 0.31043535]
Mean: 0.31286221219866617
Standard deviation: 0.011154876523176923

>>> display_scores(rf_rmse_scores)
Scores: [0.25463742 0.24316711 0.24797083 0.24010446 0.23350186 0.24951792
 0.24806436 0.23225629 0.23227359 0.25888471]
Mean: 0.24403785441735462
Standard deviation: 0.00893830076030914

 #provide date in form 19480103
 def predictVsActuals(day):
     precipPredictDay = X.loc[X['DATE'] == day]
     precipPredictValue = round(rf_model.predict(precipPredictDay)[0], 3)


     precipActual = wheather.loc[wheather['DATE'] == day]
     precipActualValue = round(precipActual.iloc[0]["PRCP"], 3)

     print("On", precipActual.iloc[0]["DATE"], "there was a high of", precipActual.iloc[0]["TMAX"], "and a low of", precipActual.iloc[0]["TMIN"])
     print("There were", precipActualValue, "inches of rainfall")
     print("We predicted",precipPredictValue, "inches of rainfall")
     print("we were off by ", round(abs(precipActualValue-precipPredictValue), 3), "inches")
     print("\n")

>>> predictVsActuals(19550302)
on 19550302 there was a high of 40 and a low of 29
There were 0.17 inches of rainfall
We predicted 0.135 inches of rainfall
we were off by  0.035 inches

>>> predictVsActuals(19880502)
on 19880502 there was a high of 47 and a low of 41
There were 0.33 inches of rainfall
We predicted 0.335 inches of rainfall
we were off by  0.005 inches


>>> predictVsActuals(19960815)
On 19960205 there was a high of 54 and a low of 40
There were 0.82 inches of rainfall
We predicted 0.021 inches of rainfall
we were off by  0.799 inches