Use fastText training classification model

import fasttext
import pandas as pd
import jieba

# Training data path
train_data_path = "./name_question.train"
 # Verify data path
valid_data_path = "./name_question.valid"

 #          
word_train_data_path = "./name_question.train.2"
word_valid_data_path = "./name_question.valid.2"


 # Read training data, separator is space
df1 = pd.read_csv(train_data_path, header=None, sep=" ")
df2 = pd.read_csv(valid_data_path, header=None, sep=" ")

 # #      
df1[1] = pd.DataFrame(map(lambda x: " ".join(jieba.cut(x)), df1[1]))
df2[1] = pd.DataFrame(map(lambda x: " ".join(jieba.cut(x)), df2[1]))


 #          
df1.to_csv(word_train_data_path, header=False, index=False, sep="\t")
 # Generate verification data
df2.to_csv(word_valid_data_path, header=False, index=False, sep="\t")

 # Take model training, here we set N-Gram features to 2
 # Other parameters are used by default, such as: Embed_size is 100, the number of training wheels EPOCH is 5
# model = fasttext.train_supervised(input=word_train_data_path, wordNgrams=2)

 # Model tuning using the automatic super-parameter tuning method
 # AutotuneValidationFile parameters Require specified path where the verification dataset is located,
 # It will use a random search method to use a random search method on the verification set.
 # Use the autotuneduration parameter to control the time of random search, the default is 300s,
 # According to different needs, we can extend or shorten the time.
model = fasttext.train_supervised(
    input=word_train_data_path,
    autotuneValidationFile=word_valid_data_path,
    autotuneDuration=600,
    wordNgrams=2,
)

 After we have verified on the verification set
valid_result = model.test(word_valid_data_path)
print(valid_result)

# Save the model
import time
time_ = int(time.time())
model_save_path = "./name_question_{}.bin".format(time_)
model.save_model(model_save_path)

# Loading model
 #                      
model = fasttext.load_model(model_save_path)
 # Predict the sample
 Result = model.predict ("" ".join (List (" Is it still classmate? "))))))
print(result)
 Result = model.predict ("" ".join (List (" surname? ")))))
print(result)