compare.py

print forecast_2w

forecast_2w[0] == b_0 + (coeff * 12) // array([ True], dtype=bool)
forecast_2w[1] == b_0 + (coeff * 13) // array([ True], dtype=bool)

compare.R

cat(forecast_2w)
forecast_2w[1] == b_0 + (coeff * 12) // TRUE
forecast_2w[2] == b_0 + (coeff * 13) // TRUE

fit-lm.py

lr = LinearRegression()
lr.fit(week, revenue)

b_0   = lr.intercept_
coeff = lr.coef_

fit-lm.R

revenue_fit <- lm(revenue ~ week, data=data)

b_0   <- revenue_fit$coefficients[[1]]
coeff <- revenue_fit$coefficients[[2]]

forecast.py

# Not query pretty, but we align our week matrices.
predict_week = np.array(
  [a for a in xrange(max(week)+1, max(week)+3)]
)[:, np.newaxis]

forecast_2w = lr.predict(predict_week)

forecast.R

# Let's forcast the next 2 weeks.
data_ahead  <- data.frame(week=c(length(data$week)+1, length(data$week)+2))
forecast_2w <- predict(revenue_fit, data_ahead)

import.py

import pandas as pd
import numpy as np
import matplotlib.pylab as plt

from sklearn.linear_model import LinearRegression
from sklearn.metrics import r2_score

predict.py

# Let's predict the values for existing weeks (Testing)
pred = lr.predict(test_week)

plt.scatter(week, rev, color='b')
plt.scatter(test_week, pred, color='red')
plt.show()

predict.R

test <- data
predict_test = predict(revenue_fit, test)

# Look at it.
plot(revenue ~ week, data=data)
lines(predict_test, col="red")

revenue-example.csv

rsquared.R

rsquared <- function(y_true, y_pred) {
  numerator   <- sum((y_true - y_pred) ** 2)
  denominator <- sum((y_true - mean(y_true)) ** 2)
  1 - numerator/denominator
}

see-example.R

data <- read.csv("/path/to/revenue-example.csv", header=TRUE)
plot(data)

setup.py

data = pd.read_csv("/path/to/revenue-example.csv", sep=",")
week     = data['week'][:, np.newaxis]
revenue  = data['revenue']

test-points.py

# Let's just test some points.
test_week = week[1:7]
test_rev  = revenue[1:7]

verify.py

# sklearn has an r2_score method. 
score = r2_score(rev, lr.predict(week[:]))
print score

# Or you can `score` the one from LinearRegression
score = lr.score(week[:], rev[:])
print score