Skip to content

A
Automate CV Processing Code
Commit 1504eea2 authored by Chamila Dilshan's avatar Chamila Dilshan
Browse files
Options

Front end

parent ec524bb8
Hide whitespace changes
Inline Side-by-side
Showing with 253 additions and 0 deletions
main.py 0 → 100644
View file @ 1504eea2
import streamlit as st
import numpy as np
import pandas as pd
import nltk
nltk.download('stopwords')
import spacy
import sys
import sklearn
import pickle
import io
from sklearn.feature_extraction.text import TfidfVectorizer
import neattext.functions as nfx
import re
import string
from pyresparser import ResumeParser
from docx import Document
from PIL import Image
import os
st.set_page_config(layout="wide")
st.markdown(
f"""
<style>
.reportview-container .main .block-container{{
max-width: 100%;
background-color: #F5F5F5;
padding-top: 5rem;
padding-right: 5rem;
padding-left: 5rem;
padding-bottom: 5rem;
}}
img{{
max-width:99%;
margin-bottom:99px;
justify-content: center;
}}
</style>
""",
unsafe_allow_html=True,
)
header = st.container()
input_file = st.container()
correlation = st.container()
feature_imp = st.container()
feature_dist = st.container()
def save_uploadedfile(uploadedfile):
with open(os.path.join("tempDir",uploadedfile.name),"wb") as f:
f.write(uploadedfile.getbuffer())
return st.success("Saved File:{} to tempDir".format(uploadedfile.name))
#Load the Model and vectorizer
loaded_model = pickle.load(open("resume_screening_model.pickle", "rb"))
loaded_vec = pickle.load(open("resume_screening_vectorizer.pickle","rb"))
with header:
st.markdown("<h1 style='text-align: center;'>Welcome To The Smart Labour Turnover Solutions' Resume Screening System.</h1>", unsafe_allow_html=True)
st.markdown("")
st.markdown("* Companies receive hundreds of resumes for their job postings, and it is very time-consuming to go through every resume.")
st.markdown("* Our System is mainly capable of automating the resume screening process using machine learning algorithms and natural language processing (NLP) techniques.")
st.markdown(""" * Following are the features of our resume screening system.
- Can input five resumes at once,
- Display the best job position for each candidate,
- Display each candidate's probability for all the job positions available,
- All the records can be downloaded in .csv format. """)
st.markdown("* You can input 5 resumes at once in .pdf, .doc, or .txt formats.")
st.markdown("* When you upload the resume, you must name the file with the candidate's name. ex: John Wick.pdf")
st.markdown("* Then our system will parse the resumes using NLP's name entity recognition (NER) techniques and display each candidate's name, skills, and experience.")
st.markdown("* Based on complete extracted data (skills, experience, degree, designation), our system will choose the best position for each candidate from 25 different job positions listed below.")
image = Image.open("position.png")
st.image(image)
# Resume files Inpuation
with input_file:
# resumes Upoload
st.subheader("1. Please Upload All Your Five Resumes in .pdf, .doc, .docx or .txt Formats")
file1 = st.file_uploader("Choose the Resume 1")
file2 = st.file_uploader("Choose the Resume 2")
file3 = st.file_uploader("Choose the Resume 3")
file4 = st.file_uploader("Choose the Resume 4")
file5 = st.file_uploader("Choose the Resume 5")
# Resume Parser
if file1 is not None:
file_details1 = {"FileName":file1.name,"FileType":file1.type}
file_name1 = file1.name[:-4]
data1 = ResumeParser(file1).get_extracted_data()
data1 = {key: data1[key] for key in data1.keys() & {'skills', 'designation', 'experience', 'degree'}}
st.markdown("<h2 style='text-align: center;'>Welcome to the Resume Parser</h2>", unsafe_allow_html=True)
st.subheader("2. Here is the Most Critical Extracted Information in the Resumes.")
st.markdown("* You can get an overview of the candidate's profile according to the below-extracted data.")
#data Cleaning and vectorization
dt1 = str(data1)
dt1 = re.sub(r'[^\w\s]', '', dt1) # remove special charcters and puctuations
dt1 = dt1.lower() # lower case conversion
dt1 = re.sub("\d+", "", dt1) # remove numbers
#do Predictions
pred1 = loaded_model.predict(loaded_vec.transform([dt1])[0])
data1["Name"] = file_name1
data1["Best_Position"] = str(pred1)
df = pd.DataFrame()
df = df.append(data1, ignore_index = True)
#pred_df
pred_val1 = loaded_model.predict_proba(loaded_vec.transform([dt1])[0])
pred_df1 = pd.DataFrame(pred_val1, columns=loaded_model.classes_, index=[file_name1])
if file2 is not None:
file_details2 = {"FileName":file2.name,"FileType":file2.type}
file_name2 = file2.name[:-4]
data2 = ResumeParser(file2).get_extracted_data()
data2 = {key: data2[key] for key in data2.keys() & {'skills', 'designation', 'experience', 'degree'}}
#data Cleaning and vectorization
dt2 = str(data2)
dt2 = re.sub(r'[^\w\s]', '', dt2) # remove special charcters and puctuations
dt2 = dt2.lower() # lower case conversion
dt2 = re.sub("\d+", "", dt2) # remove numbers
#do Predictions
pred2 = loaded_model.predict(loaded_vec.transform([dt2])[0])
data2["Name"] = file_name2
data2["Best_Position"] = str(pred2)
df = df.append(data2, ignore_index = True)
#pred_df
pred_val2 = loaded_model.predict_proba(loaded_vec.transform([dt2])[0])
pred_df2 = pd.DataFrame(pred_val2, columns=loaded_model.classes_, index=[file_name2])
if file3 is not None:
file_details3 = {"FileName":file3.name,"FileType":file3.type}
file_name3 = file3.name[:-4]
data3 = ResumeParser(file3).get_extracted_data()
data3 = {key: data3[key] for key in data3.keys() & {'skills', 'designation', 'experience', 'degree'}}
#data Cleaning and vectorization
dt3 = str(data3)
dt3 = re.sub(r'[^\w\s]', '', dt3) # remove special charcters and puctuations
dt3 = dt3.lower() # lower case conversion
dt3 = re.sub("\d+", "", dt3) # remove numbers
#do Predictions
pred3 = loaded_model.predict(loaded_vec.transform([dt3])[0])
data3["Name"] = file_name3
data3["Best_Position"] = str(pred3)
df = df.append(data3, ignore_index = True)
#pred_df
pred_val3 = loaded_model.predict_proba(loaded_vec.transform([dt3])[0])
pred_df3 = pd.DataFrame(pred_val3, columns=loaded_model.classes_, index=[file_name3])
if file4 is not None:
file_details4 = {"FileName":file4.name,"FileType":file4.type}
file_name4 = file4.name[:-4]
data4 = ResumeParser(file4).get_extracted_data()
data4 = {key: data4[key] for key in data4.keys() & {'skills', 'designation', 'experience', 'degree'}}
#data Cleaning and vectorization
dt4 = str(data4)
dt4 = re.sub(r'[^\w\s]', '', dt4) # remove special charcters and puctuations
dt4 = dt4.lower() # lower case conversion
dt4 = re.sub("\d+", "", dt4) # remove numbers
#do Predictions
pred4 = loaded_model.predict(loaded_vec.transform([dt4])[0])
data4["Name"] = file_name4
data4["Best_Position"] = str(pred4)
df = df.append(data4, ignore_index = True)
#pred_df
pred_val4 = loaded_model.predict_proba(loaded_vec.transform([dt4])[0])
pred_df4 = pd.DataFrame(pred_val4, columns=loaded_model.classes_, index=[file_name4])
if file5 is not None:
file_details5 = {"FileName":file5.name,"FileType":file5.type}
file_name5 = file5.name[:-4]
data5 = ResumeParser(file5).get_extracted_data()
data5 = {key: data5[key] for key in data5.keys() & {'skills', 'designation', 'experience', 'degree'}}
#data Cleaning and vectorization
dt5 = str(data5)
dt5 = re.sub(r'[^\w\s]', '', dt5) # remove special charcters and puctuations
dt5 = dt5.lower() # lower case conversion
dt5 = re.sub("\d+", "", dt5) # remove numbersS
#do Predictions
pred5 = loaded_model.predict(loaded_vec.transform([dt5])[0])
data5["Name"] = file_name5
data5["Best_Position"] = str(pred5)
df = df.append(data5, ignore_index = True)
#pred_df
pred_val5 = loaded_model.predict_proba(loaded_vec.transform([dt5])[0])
pred_df5 = pd.DataFrame(pred_val5, columns=loaded_model.classes_, index=[file_name5])
final_pred = pd.DataFrame(np.concatenate([pred_val1, pred_val2, pred_val3, pred_val4, pred_val5], axis=0), columns= loaded_model.classes_,index=[file_name1,file_name2,file_name3,file_name4,file_name5])
st.dataframe(data=df.loc[:, ["Name","skills","experience","degree"]])
# Downloads
st.markdown("* Please click the download button below to download the Extracted Resumes Data Report.")
@st.cache
def convert_df(df):
# IMPORTANT: Cache the conversion to prevent computation on every rerun
return df.to_csv().encode('utf-8')
csv1 = convert_df(df.loc[:, ["Name","skills","experience","degree"]])
csv2 = convert_df(df.loc[:, ["Name","Best_Position"]])
st.download_button(
label="Click here to download the Extracted Resumes Data Report",
data=csv1,
file_name='Extracted Resumes Data Report - Smart Labour Turnover Solutions.csv',
mime='text/csv',)
#Resume Classification - Part 1 - Best Position Prediction
st.markdown("<h2 style='text-align: center;'>Welcome to the Resume Classifier</h2>", unsafe_allow_html=True)
st.subheader("3. Here is the Best-Predicted Position for Each Candidate.")
st.markdown("* Below is the best position for each candidate according to their skills, experience, degree, and designation.")
col1, col2, col3 = st.columns(3)
with col1:
st.write("")
with col2:
st.dataframe(data=df.loc[:, ["Name","Best_Position"]])
with col3:
st.write("")
st.write("")
#downloads
st.markdown("* Please click the download button below to download the Best Position Prediction Report.")
st.download_button(
label="Click here to download the Best Position Prediction Report",
data=csv2,
file_name='Best Podition Prediction Report - Smart Labour Turnover Solutions.csv',
mime='text/csv',)
#Resume Classification - Part 2 - All Positions Probability
st.subheader("4. Here is Each Candidate's Probability for All 25 Job Positions Available.")
st.markdown("* By examining each candidate's probability for each job position, you can select the best candidates even if they are suitable for the same job position. ")
pred1, pred2, pred3, pred4, pred5 = st.columns(5)
with pred1:
st.dataframe(data=pred_df1[pred_df1.iloc[-1].sort_values(ascending=False).index].T, height=914)
with pred2:
st.dataframe(data=pred_df2[pred_df2.iloc[-1].sort_values(ascending=False).index].T, height=914)
with pred3:
st.dataframe(data=pred_df3[pred_df3.iloc[-1].sort_values(ascending=False).index].T, height=914)
with pred4:
st.dataframe(data=pred_df4[pred_df4.iloc[-1].sort_values(ascending=False).index].T, height=914)
with pred5:
st.dataframe(data=pred_df5[pred_df5.iloc[-1].sort_values(ascending=False).index].T, height=914)
#downloads
#downloads
st.markdown("* Please click the download button below to download the All positions Prediction Report.")
@st.cache
def convert_df(df):
# IMPORTANT: Cache the conversion to prevent computation on every rerun
return df.to_csv().encode('utf-8')
csv3 = convert_df(final_pred)
st.download_button(
label="Click here to download the All positions Prediction Report",
data=csv3,
file_name='All positions Prediction Report - Smart Labour Turnover Solutions.csv',
mime='text/csv',)
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment