How to Build a Python Machine Learning Environment and Fit Your First Model

This article explains how to set up a Python machine learning environment using Python 2.7 and the open‑source tools NumPy, SciPy, Matplotlib, and scikit‑learn.

First, install Python 2.7, then install the required packages (NumPy, SciPy, Matplotlib) and scikit‑learn 0.15, all of which can be downloaded from the provided Baidu Cloud links.

Next, read the web traffic data (a TSV file with 743 rows) using scipy.genfromtxt:

import scipy as sp

data = sp.genfromtxt("web_traffic.tsv", delimiter="\t")

Inspect the shape with print(data.shape) and notice that some rows contain nan values.

Remove invalid rows by filtering out NaNs:

x = data[:,0]

y = data[:,1]

x = x[~sp.isnan(y)]

y = y[~sp.isnan(y)]

Visualize the cleaned data with Matplotlib:

import matplotlib.pyplot as plt

plt.scatter(x, y)

plt.title("Web traffic over the last month")

plt.xlabel("Time")

plt.ylabel("Hits/hour")

plt.show()

Define an error function for regression:

def error(f, x, y):

return sp.sum((f(x) - y) ** 2)

Fit a first‑order polynomial (a straight line) to the data using polyfit:

fp1, residuals, rank, sv, rcond = sp.polyfit(x, y, 1, full=True)

Print the residuals to see the fitting error and the resulting model: print(residuals) The fitted line is f(x) = 2.59619213 * x + 989.02487106.

Create a callable model with poly1d and plot the fitted line together with the data:

f1 = sp.poly1d(fp1)

fx = sp.linspace(0, x[-1], 1000)

plt.plot(fx, f1(fx), linewidth=4)

plt.legend(["d=%i" % f1.order], loc="upper left")

plt.show()

The linear fit is clearly insufficient, illustrating the need for more advanced algorithms, which scikit‑learn provides.