Recognition and classification of handwritten digits + technical summary

(1) Learn to reprint an article on machine learning handwritten digit recognition

Python 3 uses machine learning model for handwritten digit detection

Python 3 generates a dataset of handwritten digits

(2) Technical summary

In the initial stage of machine learning code implementation, it is necessary to start the project by yourself, but also to read more blog posts of other predecessors, especially some projects with additional algorithm ideas. Do it yourself, only in the process of code implementation can you find many problems, summarize the basic code, and spare.

Folders are automatically generated:

@author: Wang Gan
"""
import them
path_folders = "H:/Python data/own code/handwritten number recognition/P_generate_handwritten_number/data_pngs/"

# 1-9
for i in range(49,58):
    if (os.path.isdir(path_folders + chr(i))):
        pass
    else:
        # print(i,": ",path_1+chr(i))
        # build directory
        os.mkdir(path_folders+chr(i))

Renaming of files:

# Rename CSV with "test_+samples.csv"
  if "data_"+str(sample_nums)+".csv" in tmpdir:
        # Data_XXX.csv has been generated before and needs to be deleted first
        os.remove(path_csv+"data_"+str(sample_nums)+".csv")
        os.rename(path_csv+"test1.csv", path_csv+"data_"+str(sample_nums)+".csv")
  else:          os.rename(path_csv+"test1.csv", path_csv+"data_"+str(sample_nums)+".csv")

Sinicization path pandas open:

# read CSV data
column_names = ["samples", "acc_LR", "acc_LSVC", "acc_MLPC", "acc_SGDC"]
f=open('H:/Python data/own code/handwritten digit recognition/Feature_extract/data/score_csv/'+'score.csv')
rd_csv = pd.read_csv( f , names=column_names)

How to call a class in one file from another file

Knowledge about python main function

The accuracy is not high enough Reason analysis: (The reason is unknown)

                               My own results provided by the original author

Problem solved: next blog post.