Python Linear Regression: I cant use dates as my independent variable properly

Question

I wanted to predict the number of customers for the next 7 days. So Im going to use Linear Regression with Dates as my independent variable and # of Customers as my dependent variable. But I get weird values in the Coefficient and Intercept, that I doubt will result in good predictions.

This is my dataset from the csv file:

date,customers
1/1/2022,13
1/2/2022,23
1/3/2022,15
1/4/2022,45
1/5/2022,89
1/6/2022,34
1/7/2022,53

Here is my code:

data = pd.read_csv("data/Customers.csv", converters={"date": pd.Timestamp})

plt.scatter(data.date, data.customers)
plt.show()

linreg = linear_model.LinearRegression()
linreg.fit(data[['date']], data.customers)

#Coefficient
print(linreg.coef_)
#Intercept
print(linreg.intercept_)
#Prediction
print(linreg.coef_ * data.date[0].toordinal()) + linreg.intercept_)

These are the results:

Coefficient = [8.92857143e-14]
Intercept = -146501.71428571438
Prediction = [-146501.71428565]

if you think the dates are at fault - replace them with 0,1,2,3,4, ... and try again. — Patrick Artner, Feb 09 '22 at 13:40
I see much better numbers when I use JUST NUMBERS such as Coef = [7.71428571] , Intercept = 15.714285714285712, and Prediction = [69.71428571] — Ron Ancheta, Feb 09 '22 at 13:45
https://stackoverflow.com/questions/37292872/how-can-i-one-hot-encode-in-python — Patrick Artner, Feb 09 '22 at 14:06
Im not yet familiar with One Hot Encoding. Would you mind explaining how it would work on my code and how it would solve the problem? — Ron Ancheta, Feb 09 '22 at 14:27

Python Linear Regression: I cant use dates as my independent variable properly

0 Answers0