How can I return accuracy rates for Top N predictions using sklearn's SGDClassifier?

问题

I am trying to modify the results in this post (How to get Top 3 or Top N predictions using sklearn's SGDClassifier) to get the accuracy rate returned, however I am get an accuracy rate of zero and I can't figure out why. Any thoughts? Any thoughts/edits would be much appreciated! Thank you.

from sklearn.feature_extraction.text import TfidfVectorizer
import numpy as np
from sklearn import linear_model
arr=['dogs cats lions','apple pineapple orange','water fire earth air', 'sodium potassium calcium']
vectorizer = TfidfVectorizer()
X = vectorizer.fit_transform(arr)
feature_names = vectorizer.get_feature_names()
Y = ['animals', 'fruits', 'elements','chemicals']
T=["eating apple roasted in fire and enjoying fresh air"]
test = vectorizer.transform(T)
clf = linear_model.SGDClassifier(loss='log')
clf.fit(X,Y)
x=clf.predict(test)

def top_n_accuracy(probs, test, n):
    best_n = np.argsort(probs, axis=1)[:,-n:]
    ts = np.argmax(test, axis=1)
    successes = 0
    for i in range(ts.shape[0]):
        if ts[i] in best_n[i,:]:
            successes += 1
    return float(successes)/ts.shape[0]

n=2
probs = clf.predict_proba(test)

top_n_accuracy(probs, test, n)

回答1:

from sklearn.feature_extraction.text import TfidfVectorizer
import numpy as np
from sklearn import linear_model

arr=['dogs cats lions','apple pineapple orange','water fire earth air', 'sodium potassium calcium']
vectorizer = TfidfVectorizer()
X = vectorizer.fit_transform(arr)
feature_names = vectorizer.get_feature_names()
Y = ['animals', 'fruits', 'elements','chemicals']
T=["eating apple roasted in fire and enjoying fresh air", "I love orange"]
test = vectorizer.transform(T)

clf = linear_model.SGDClassifier(loss='log')
clf.fit(X,Y)
x=clf.predict(test)

n=2
probs = clf.predict_proba(test)

topn = np.argsort(probs, axis = 1)[:,-n:]

Here I introduce the ground truth label vector (these are numeric indices, you need to map ["elements", etc] to [0,1,2 etc]. Here I assumed your test example belongs to elements.

y_true = np.array([2,1])

This should then compute your accuracy

np.mean(np.array([1 if y_true[k] in topn[k] else 0 for k in range(len(topn))]))

回答2:

I ended up figuring this on out, albeit a bit different than the above...

# Set Data Location:
data = 'top10000.csv'  

# load the data
df = pd.read_csv(data,low_memory=False,thousands=',', encoding='latin-1')
df = df.dropna()
df = df[['CODE','DUTIES']] #select only these columns
#df = df.rename(index=float, columns={"CODE": "label", "DUTIES": "text"})
df = df.rename(columns={"CODE": "label", "DUTIES": "text"})

#Convert label to float so you don't need to encode for processing later on
df['label']=df['label'].str.replace('-', '',regex=True, case = False).str.strip()
df['label']=df['label'].str.replace('.', '',regex=True)
#df['label']=pd.to_numeric(df['label'])
df['label']=df['label'].str[1:].astype(int)
#df['label'].astype('float64', raise_on_error = True)

#split data into testing and training
train_x, valid_x, train_y, valid_y = model_selection.train_test_split(df.text, df.label,test_size=0.33, random_state=6)

#reset the index 
valid_y = valid_y.reset_index(drop=True)
valid_x = valid_x.reset_index(drop=True)

# We will also copy the validation datasets to a dataframe to be able to merge later on
valid_x_df = pd.DataFrame(valid_x)
valid_y_df = pd.DataFrame(valid_y)

# Extracte features 
count_vect = CountVectorizer()
X_train_counts = count_vect.fit_transform(train_x)
X_test_counts = count_vect.transform(valid_x)

# Define the model training and validation function
def TV_model(classifier, feature_vector_train, label, feature_vector_valid, valid_y, valid_x, is_neural_net=False):

    # fit the training dataset on the classifier
    classifier.fit(feature_vector_train, label)

    # predict the top n labels on validation dataset
    n = 5
    #classifier.probability = True
    probas = classifier.predict_proba(feature_vector_valid)
    predictions = classifier.predict(feature_vector_valid)

    #Identify the indexes of the top predictions
    top_n_predictions = np.argsort(probas, axis = 1)[:,-n:]

    #then find the associated SOC code for each prediction
    top_class = classifier.classes_[top_n_predictions]

    #cast to a new dataframe
    top_class_df = pd.DataFrame(data=top_class)

    #merge it up with the validation labels and descriptions
    results = pd.merge(valid_y, valid_x, left_index=True, right_index=True)
    results = pd.merge(results, top_class_df, left_index=True, right_index=True)

    # Top 5 results condiions and choices
    top5_conditions = [
        (results.iloc[:,0] == results[0]),
        (results.iloc[:,0] == results[1]),
        (results.iloc[:,0] == results[2]),
        (results.iloc[:,0] == results[3]),
        (results.iloc[:,0] == results[4])]
    top5_choices = [1, 1, 1, 1, 1]

    # Fetch Top 1 Result
    top1_conditions = [(results.iloc[:,0] == results[4])]
    top1_choices = [1]

    # Create the success columns
    results['Top 5 Successes'] = np.select(top5_conditions, top5_choices, default=0)
    results['Top 1 Successes'] = np.select(top1_conditions, top1_choices, default=0)

    #Print the QA 
    print("Are Top 5 Results greater than Top 1 Result? (answer must be True): ", (sum(results['Top 5 Successes'])/results.shape[0])>(metrics.accuracy_score(valid_y, predictions)))
    print("Are Top 1 Results equal from predict() and predict_proba()? (answer must be True): ", (sum(results['Top 1 Successes'])/results.shape[0])==(metrics.accuracy_score(valid_y, predictions)))
    print(" ")
    print("Details: ")
    print("Top 5 Accuracy Rate (predict_proba)= ", sum(results['Top 5 Successes'])/results.shape[0])
    #print("Top 5 Accuracy Rate (np.mean)= ", np.mean(np.array([1 if valid_y[k] in top_class[k] else 0 for k in range(len(top_class))])))
    print("Top 1 Accuracy Rate (predict_proba)= ", sum(results['Top 1 Successes'])/results.shape[0])
    print("Top 1 Accuracy Rate = (predict)", metrics.accuracy_score(valid_y, predictions))

# Train and validate model from example data using the function defined above
TV_model(LogisticRegression(), X_train_counts, train_y, X_test_counts, valid_y_df, valid_x_df)

I'm sure it could be more computationally efficient, so any suggestion on how I could transform the accuracy rate calculation into a one liner like was suggested in the comments above would be much appreciated!

来源：https://stackoverflow.com/questions/54855656/how-can-i-return-accuracy-rates-for-top-n-predictions-using-sklearns-sgdclassif