Estimating the income bracket_Python Machine Learning Cookbook-QQ阅读男频武侠网

上QQ阅读APP看书，第一时间看更新

Estimating the income bracket

We will build a classifier to estimate the income bracket of a person based on 14 attributes. The possible output classes are higher than 50K or lower than or equal to 50K. There is a slight twist in this dataset in the sense that each datapoint is a mixture of numbers and strings. Numerical data is valuable, and we cannot use a label encoder in these situations. We need to design a system that can deal with numerical and non-numerical data at the same time. We will use the census income dataset available at https://archive.ics.uci.edu/ml/datasets/Census+Income.

How to do it…

We will use the income.py file already provided to you as a reference. We will use a Naive Bayes classifier to achieve this. Let's import a couple of packages:
```
from sklearn import preprocessing
from sklearn.naive_bayes import GaussianNB 
```

Let's load the dataset:

input_file = 'path/to/adult.data.txt'

# Reading the data
X = []
y = []
count_lessthan50k = 0
count_morethan50k = 0
num_images_threshold = 10000

We will use 20,000 datapoints from the datasets—10,000 for each class to avoid class imbalance. During training, if you use many datapoints that belong to a single class, the classifier tends to get biased toward this class. Therefore, it's better to use the same number of datapoints for each class:

with open(input_file, 'r') as f:
    for line in f.readlines():
        if '?' in line:
            continue

        data = line[:-1].split(', ')

        if data[-1] == '<=50K' and count_lessthan50k < num_images_threshold:
            X.append(data)
            count_lessthan50k = count_lessthan50k + 1

        elif data[-1] == '>50K' and count_morethan50k < num_images_threshold:
            X.append(data)
            count_morethan50k = count_morethan50k + 1

        if count_lessthan50k >= num_images_threshold and count_morethan50k >= num_images_threshold:
            break

X = np.array(X)

It's a comma-separated file again. We just loaded the data in the X variable just like before.

We need to convert string attributes to numerical data while leaving out the original numerical data:

# Convert string data to numerical data
label_encoder = [] 
X_encoded = np.empty(X.shape)
for i,item in enumerate(X[0]):
    if item.isdigit():
        X_encoded[:, i] = X[:, i]
    else: 
        label_encoder.append(preprocessing.LabelEncoder())
        X_encoded[:, i] = label_encoder[-1].fit_transform(X[:, i])

X = X_encoded[:, :-1].astype(int)
y = X_encoded[:, -1].astype(int)

The isdigit()function helps us in identifying numerical data. We converted string data to numerical data and stored all the label encoders in a list so that we can use it when we want to classify unknown data.

Let's train the classifier:

# Build a classifier
classifier_gaussiannb = GaussianNB()
classifier_gaussiannb.fit(X, y)

Let's split the data into training and testing to extract performance metrics:

# Cross validation
from sklearn import cross_validation

X_train, X_test, y_train, y_test = cross_validation.train_test_split(X, y, test_size=0.25, random_state=5)
classifier_gaussiannb = GaussianNB()
classifier_gaussiannb.fit(X_train, y_train)
y_test_pred = classifier_gaussiannb.predict(X_test)

Let's extract performance metrics:

# compute F1 score of the classifier
f1 = cross_validation.cross_val_score(classifier_gaussiannb, 
        X, y, scoring='f1_weighted', cv=5)
print "F1 score: " + str(round(100*f1.mean(), 2)) + "%"

Let's see how to classify a single datapoint. We need to convert the datapoint into something that our classifier can understand:

# Testing encoding on single data instance
input_data = ['39', 'State-gov', '77516', 'Bachelors', '13', 'Never-married', 'Adm-clerical', 'Not-in-family', 'White', 'Male', '2174', '0', '40', 'United-States'] 
count = 0
input_data_encoded = [-1] * len(input_data)
for i,item in enumerate(input_data):
    if item.isdigit():
        input_data_encoded[i] = int(input_data[i])
    else:
        input_data_encoded[i] = int(label_encoder[count].transform(input_data[i]))
        count = count + 1 

input_data_encoded = np.array(input_data_encoded)

We are now ready to classify it:

# Predict and print output for a particular datapoint
output_class = classifier_gaussiannb.predict(input_data_encoded)
print label_encoder[-1].inverse_transform(output_class)[0]

Just like before, we use the predict method to get the output class and the inverse_transform method to convert this label back to its original form to print it out on the Terminal.