Now, since we have cleaned the bikesData data set, let us split it into Training and Test data sets into 70:30 ratio using scikit-learn's train_test_split() function.
Also, train_test_split() function uses 'Random Sampling', hence resulting train_set and test_set data sets have to be sorted by dayCount.
Random Sampling may not be the best way to split the data, what other types of best Sampling method you can think of?
We will also define an utility function named display_scores. This function is used to calculate the basics stats of observed scores from cross-validation of models. Please copy this function in your code, we will be using it often in this project.
Set np random seed to 42 using code below to ensure the results of the exercise are repeatable.
np.random.seed(42)
Import train_test_split function from scikit-learn's model_selection
Please add a new feature(column) dayCount to bikesData data set using below code:
bikesData['dayCount'] = pd.Series(range(bikesData.shape[0]))/24
Split the bikesData data set into Training set train_set and Test set test_set in 70:30 ratio using scikit-learn's train_test_split() function.
Sort the train_set and test_set values by dayCount by using the below code:
train_set.sort_values('dayCount', axis= 0, inplace=True)
test_set.sort_values('dayCount', axis= 0, inplace=True)
Now print the 'number of instances' for train_set and test_set data sets.
Finally, create the function display_scores as shown below:
def display_scores(scores):
print("Scores:", scores)
print("Mean:", scores.mean())
print("Standard deviation:", scores.std())
Loading comments...