Chapter 1 – The Machine Learning landscape
This is the code used to generate some of the figures in chapter 1.
Run in Google ColabCode example 1-1¶
Although Python 2.x may work, it is deprecated so we strongly recommend you use Python 3 instead.
In [1]:
# Python ≥3.5 is required
import sys
assert sys.version_info >= (3, 5)
In [2]:
# Scikit-Learn ≥0.20 is required
import sklearn
assert sklearn.__version__ >= "0.20"
This function just merges the OECD's life satisfaction data and the IMF's GDP per capita data. It's a bit too long and boring and it's not specific to Machine Learning, which is why I left it out of the book.
In [3]:
def prepare_country_stats(oecd_bli, gdp_per_capita):
oecd_bli = oecd_bli[oecd_bli["INEQUALITY"]=="TOT"]
oecd_bli = oecd_bli.pivot(index="Country", columns="Indicator", values="Value")
gdp_per_capita.rename(columns={"2015": "GDP per capita"}, inplace=True)
gdp_per_capita.set_index("Country", inplace=True)
full_country_stats = pd.merge(left=oecd_bli, right=gdp_per_capita,
left_index=True, right_index=True)
full_country_stats.sort_values(by="GDP per capita", inplace=True)
remove_indices = [0, 1, 6, 8, 33, 34, 35]
keep_indices = list(set(range(36)) - set(remove_indices))
return full_country_stats[["GDP per capita", 'Life satisfaction']].iloc[keep_indices]
The code in the book expects the data files to be located in the current directory. I just tweaked it here to fetch the files in datasets/lifesat.
In [4]:
import os
datapath = os.path.join("datasets", "lifesat", "")
In [5]:
# To plot pretty figures directly within Jupyter
%matplotlib inline
import matplotlib as mpl
mpl.rc('axes', labelsize=14)
mpl.rc('xtick', labelsize=12)
mpl.rc('ytick', labelsize=12)
In [6]:
# Download the data
import urllib.request
DOWNLOAD_ROOT = "https://raw.githubusercontent.com/ageron/handson-ml2/master/"
os.makedirs(datapath, exist_ok=True)
for filename in ("oecd_bli_2015.csv", "gdp_per_capita.csv"):
print("Downloading", filename)
url = DOWNLOAD_ROOT + "datasets/lifesat/" + filename
urllib.request.urlretrieve(url, datapath + filename)
In [7]:
# Code example
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import sklearn.linear_model
# Load the data
oecd_bli = pd.read_csv(datapath + "oecd_bli_2015.csv", thousands=',')
gdp_per_capita = pd.read_csv(datapath + "gdp_per_capita.csv",thousands=',',delimiter='\t',
encoding='latin1', na_values="n/a")
# Prepare the data
country_stats = prepare_country_stats(oecd_bli, gdp_per_capita)
X = np.c_[country_stats["GDP per capita"]]
y = np.c_[country_stats["Life satisfaction"]]
# Visualize the data
country_stats.plot(kind='scatter', x="GDP per capita", y='Life satisfaction')
plt.show()
# Select a linear model
model = sklearn.linear_model.LinearRegression()
# Train the model
model.fit(X, y)
# Make a prediction for Cyprus
X_new = [[22587]] # Cyprus' GDP per capita
print(model.predict(X_new)) # outputs [[ 5.96242338]]
In [ ]:
In [ ]:
In [ ]:
Note: you can ignore the rest of this notebook, it just generates many of the figures in chapter 1.¶
In [ ]:
In [ ]:
In [ ]:
Create a function to save the figures.
In [8]:
# Where to save the figures
PROJECT_ROOT_DIR = "."
CHAPTER_ID = "fundamentals"
IMAGES_PATH = os.path.join(PROJECT_ROOT_DIR, "images", CHAPTER_ID)
os.makedirs(IMAGES_PATH, exist_ok=True)
def save_fig(fig_id, tight_layout=True, fig_extension="png", resolution=300):
path = os.path.join(IMAGES_PATH, fig_id + "." + fig_extension)
print("Saving figure", fig_id)
if tight_layout:
plt.tight_layout()
plt.savefig(path, format=fig_extension, dpi=resolution)
Make this notebook's output stable across runs:
In [9]:
np.random.seed(42)
Load and prepare Life satisfaction data¶
If you want, you can get fresh data from the OECD's website.
Download the CSV from http://stats.oecd.org/index.aspx?DataSetCode=BLI
and save it to datasets/lifesat/.
In [10]:
oecd_bli = pd.read_csv(datapath + "oecd_bli_2015.csv", thousands=',')
oecd_bli = oecd_bli[oecd_bli["INEQUALITY"]=="TOT"]
oecd_bli = oecd_bli.pivot(index="Country", columns="Indicator", values="Value")
oecd_bli.head(2)
Out[10]:
In [11]:
oecd_bli["Life satisfaction"].head()
Out[11]:
Load and prepare GDP per capita data¶
Just like above, you can update the GDP per capita data if you want. Just download data from http://goo.gl/j1MSKe (=> imf.org) and save it to datasets/lifesat/.
In [12]:
gdp_per_capita = pd.read_csv(datapath+"gdp_per_capita.csv", thousands=',', delimiter='\t',
encoding='latin1', na_values="n/a")
gdp_per_capita.rename(columns={"2015": "GDP per capita"}, inplace=True)
gdp_per_capita.set_index("Country", inplace=True)
gdp_per_capita.head(2)
Out[12]:
In [13]:
full_country_stats = pd.merge(left=oecd_bli, right=gdp_per_capita, left_index=True, right_index=True)
full_country_stats.sort_values(by="GDP per capita", inplace=True)
full_country_stats
Out[13]:
In [14]:
full_country_stats[["GDP per capita", 'Life satisfaction']].loc["United States"]
Out[14]:
In [15]:
remove_indices = [0, 1, 6, 8, 33, 34, 35]
keep_indices = list(set(range(36)) - set(remove_indices))
sample_data = full_country_stats[["GDP per capita", 'Life satisfaction']].iloc[keep_indices]
missing_data = full_country_stats[["GDP per capita", 'Life satisfaction']].iloc[remove_indices]
In [16]:
sample_data.plot(kind='scatter', x="GDP per capita", y='Life satisfaction', figsize=(5,3))
plt.axis([0, 60000, 0, 10])
position_text = {
"Hungary": (5000, 1),
"Korea": (18000, 1.7),
"France": (29000, 2.4),
"Australia": (40000, 3.0),
"United States": (52000, 3.8),
}
for country, pos_text in position_text.items():
pos_data_x, pos_data_y = sample_data.loc[country]
country = "U.S." if country == "United States" else country
plt.annotate(country, xy=(pos_data_x, pos_data_y), xytext=pos_text,
arrowprops=dict(facecolor='black', width=0.5, shrink=0.1, headwidth=5))
plt.plot(pos_data_x, pos_data_y, "ro")
plt.xlabel("GDP per capita (USD)")
save_fig('money_happy_scatterplot')
plt.show()
In [17]:
sample_data.to_csv(os.path.join("datasets", "lifesat", "lifesat.csv"))
In [18]:
sample_data.loc[list(position_text.keys())]
Out[18]:
In [19]:
import numpy as np
sample_data.plot(kind='scatter', x="GDP per capita", y='Life satisfaction', figsize=(5,3))
plt.xlabel("GDP per capita (USD)")
plt.axis([0, 60000, 0, 10])
X=np.linspace(0, 60000, 1000)
plt.plot(X, 2*X/100000, "r")
plt.text(40000, 2.7, r"$\theta_0 = 0$", fontsize=14, color="r")
plt.text(40000, 1.8, r"$\theta_1 = 2 \times 10^{-5}$", fontsize=14, color="r")
plt.plot(X, 8 - 5*X/100000, "g")
plt.text(5000, 9.1, r"$\theta_0 = 8$", fontsize=14, color="g")
plt.text(5000, 8.2, r"$\theta_1 = -5 \times 10^{-5}$", fontsize=14, color="g")
plt.plot(X, 4 + 5*X/100000, "b")
plt.text(5000, 3.5, r"$\theta_0 = 4$", fontsize=14, color="b")
plt.text(5000, 2.6, r"$\theta_1 = 5 \times 10^{-5}$", fontsize=14, color="b")
save_fig('tweaking_model_params_plot')
plt.show()
In [20]:
from sklearn import linear_model
lin1 = linear_model.LinearRegression()
Xsample = np.c_[sample_data["GDP per capita"]]
ysample = np.c_[sample_data["Life satisfaction"]]
lin1.fit(Xsample, ysample)
t0, t1 = lin1.intercept_[0], lin1.coef_[0][0]
t0, t1
Out[20]:
In [21]:
sample_data.plot(kind='scatter', x="GDP per capita", y='Life satisfaction', figsize=(5,3))
plt.xlabel("GDP per capita (USD)")
plt.axis([0, 60000, 0, 10])
X=np.linspace(0, 60000, 1000)
plt.plot(X, t0 + t1*X, "b")
plt.text(5000, 3.1, r"$\theta_0 = 4.85$", fontsize=14, color="b")
plt.text(5000, 2.2, r"$\theta_1 = 4.91 \times 10^{-5}$", fontsize=14, color="b")
save_fig('best_fit_model_plot')
plt.show()
In [22]:
cyprus_gdp_per_capita = gdp_per_capita.loc["Cyprus"]["GDP per capita"]
print(cyprus_gdp_per_capita)
cyprus_predicted_life_satisfaction = lin1.predict([[cyprus_gdp_per_capita]])[0][0]
cyprus_predicted_life_satisfaction
Out[22]:
In [23]:
sample_data.plot(kind='scatter', x="GDP per capita", y='Life satisfaction', figsize=(5,3), s=1)
plt.xlabel("GDP per capita (USD)")
X=np.linspace(0, 60000, 1000)
plt.plot(X, t0 + t1*X, "b")
plt.axis([0, 60000, 0, 10])
plt.text(5000, 7.5, r"$\theta_0 = 4.85$", fontsize=14, color="b")
plt.text(5000, 6.6, r"$\theta_1 = 4.91 \times 10^{-5}$", fontsize=14, color="b")
plt.plot([cyprus_gdp_per_capita, cyprus_gdp_per_capita], [0, cyprus_predicted_life_satisfaction], "r--")
plt.text(25000, 5.0, r"Prediction = 5.96", fontsize=14, color="b")
plt.plot(cyprus_gdp_per_capita, cyprus_predicted_life_satisfaction, "ro")
save_fig('cyprus_prediction_plot')
plt.show()
In [24]:
sample_data[7:10]
Out[24]:
In [25]:
(5.1+5.7+6.5)/3
Out[25]:
In [26]:
backup = oecd_bli, gdp_per_capita
def prepare_country_stats(oecd_bli, gdp_per_capita):
oecd_bli = oecd_bli[oecd_bli["INEQUALITY"]=="TOT"]
oecd_bli = oecd_bli.pivot(index="Country", columns="Indicator", values="Value")
gdp_per_capita.rename(columns={"2015": "GDP per capita"}, inplace=True)
gdp_per_capita.set_index("Country", inplace=True)
full_country_stats = pd.merge(left=oecd_bli, right=gdp_per_capita,
left_index=True, right_index=True)
full_country_stats.sort_values(by="GDP per capita", inplace=True)
remove_indices = [0, 1, 6, 8, 33, 34, 35]
keep_indices = list(set(range(36)) - set(remove_indices))
return full_country_stats[["GDP per capita", 'Life satisfaction']].iloc[keep_indices]
In [27]:
# Code example
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import sklearn.linear_model
# Load the data
oecd_bli = pd.read_csv(datapath + "oecd_bli_2015.csv", thousands=',')
gdp_per_capita = pd.read_csv(datapath + "gdp_per_capita.csv",thousands=',',delimiter='\t',
encoding='latin1', na_values="n/a")
# Prepare the data
country_stats = prepare_country_stats(oecd_bli, gdp_per_capita)
X = np.c_[country_stats["GDP per capita"]]
y = np.c_[country_stats["Life satisfaction"]]
# Visualize the data
country_stats.plot(kind='scatter', x="GDP per capita", y='Life satisfaction')
plt.show()
# Select a linear model
model = sklearn.linear_model.LinearRegression()
# Train the model
model.fit(X, y)
# Make a prediction for Cyprus
X_new = [[22587]] # Cyprus' GDP per capita
print(model.predict(X_new)) # outputs [[ 5.96242338]]
In [28]:
oecd_bli, gdp_per_capita = backup
In [29]:
missing_data
Out[29]:
In [30]:
position_text2 = {
"Brazil": (1000, 9.0),
"Mexico": (11000, 9.0),
"Chile": (25000, 9.0),
"Czech Republic": (35000, 9.0),
"Norway": (60000, 3),
"Switzerland": (72000, 3.0),
"Luxembourg": (90000, 3.0),
}
In [31]:
sample_data.plot(kind='scatter', x="GDP per capita", y='Life satisfaction', figsize=(8,3))
plt.axis([0, 110000, 0, 10])
for country, pos_text in position_text2.items():
pos_data_x, pos_data_y = missing_data.loc[country]
plt.annotate(country, xy=(pos_data_x, pos_data_y), xytext=pos_text,
arrowprops=dict(facecolor='black', width=0.5, shrink=0.1, headwidth=5))
plt.plot(pos_data_x, pos_data_y, "rs")
X=np.linspace(0, 110000, 1000)
plt.plot(X, t0 + t1*X, "b:")
lin_reg_full = linear_model.LinearRegression()
Xfull = np.c_[full_country_stats["GDP per capita"]]
yfull = np.c_[full_country_stats["Life satisfaction"]]
lin_reg_full.fit(Xfull, yfull)
t0full, t1full = lin_reg_full.intercept_[0], lin_reg_full.coef_[0][0]
X = np.linspace(0, 110000, 1000)
plt.plot(X, t0full + t1full * X, "k")
plt.xlabel("GDP per capita (USD)")
save_fig('representative_training_data_scatterplot')
plt.show()
In [32]:
full_country_stats.plot(kind='scatter', x="GDP per capita", y='Life satisfaction', figsize=(8,3))
plt.axis([0, 110000, 0, 10])
from sklearn import preprocessing
from sklearn import pipeline
poly = preprocessing.PolynomialFeatures(degree=60, include_bias=False)
scaler = preprocessing.StandardScaler()
lin_reg2 = linear_model.LinearRegression()
pipeline_reg = pipeline.Pipeline([('poly', poly), ('scal', scaler), ('lin', lin_reg2)])
pipeline_reg.fit(Xfull, yfull)
curve = pipeline_reg.predict(X[:, np.newaxis])
plt.plot(X, curve)
plt.xlabel("GDP per capita (USD)")
save_fig('overfitting_model_plot')
plt.show()
In [33]:
full_country_stats.loc[[c for c in full_country_stats.index if "W" in c.upper()]]["Life satisfaction"]
Out[33]:
In [34]:
gdp_per_capita.loc[[c for c in gdp_per_capita.index if "W" in c.upper()]].head()
Out[34]:
In [35]:
plt.figure(figsize=(8,3))
plt.xlabel("GDP per capita")
plt.ylabel('Life satisfaction')
plt.plot(list(sample_data["GDP per capita"]), list(sample_data["Life satisfaction"]), "bo")
plt.plot(list(missing_data["GDP per capita"]), list(missing_data["Life satisfaction"]), "rs")
X = np.linspace(0, 110000, 1000)
plt.plot(X, t0full + t1full * X, "r--", label="Linear model on all data")
plt.plot(X, t0 + t1*X, "b:", label="Linear model on partial data")
ridge = linear_model.Ridge(alpha=10**9.5)
Xsample = np.c_[sample_data["GDP per capita"]]
ysample = np.c_[sample_data["Life satisfaction"]]
ridge.fit(Xsample, ysample)
t0ridge, t1ridge = ridge.intercept_[0], ridge.coef_[0][0]
plt.plot(X, t0ridge + t1ridge * X, "b", label="Regularized linear model on partial data")
plt.legend(loc="lower right")
plt.axis([0, 110000, 0, 10])
plt.xlabel("GDP per capita (USD)")
save_fig('ridge_model_plot')
plt.show()
In [36]:
backup = oecd_bli, gdp_per_capita
def prepare_country_stats(oecd_bli, gdp_per_capita):
return sample_data
In [37]:
# Replace this linear model:
import sklearn.linear_model
model = sklearn.linear_model.LinearRegression()
In [38]:
# with this k-neighbors regression model:
import sklearn.neighbors
model = sklearn.neighbors.KNeighborsRegressor(n_neighbors=3)
In [39]:
X = np.c_[country_stats["GDP per capita"]]
y = np.c_[country_stats["Life satisfaction"]]
# Train the model
model.fit(X, y)
# Make a prediction for Cyprus
X_new = np.array([[22587.0]]) # Cyprus' GDP per capita
print(model.predict(X_new)) # outputs [[ 5.76666667]]
In [ ]: