Languages
Pages
Legal
Raw Data
Analysis
Par/cipants Your text would go here.
A Personalized Company Recommender System for Job Seekers Ruixi Lin, Yue Kang, Yixin Cai
Algorithm F1 Score Decision Tree 50.94% Naive Bayes 58.49%
linear SVM(1-v-rest) 63.52% linear SVM(1-v-1) 62.26% linear SVM(ecoc) 63.52% Neural Network 66.04%
Confusion Matrix test set
Company Precision Recall F1 score Google 56.25% 84.91% 67.67%
Facebook 80.00% 37.74% 51.28% Apple 74.07% 75.47% 74.77%
train set Company Precision Recall F1 score Google 56.19% 72.67% 63.37%
Facebook 77.68% 58.00% 66.41% Apple 64.58% 62.00% 63.27%
Results v >=1 Year(s) of Experience v >=5 Year(s) of Experience v >=10 Year(s) of Experience v >=1 Year(s) in Current Company
v >=5 Year(s) in Current Company
v Has Doctorate Degree v Has Masters Degree v Is bilingual v Has PublicaGons or Patents v >= 20 Number of Skills v Google Intern v Facebook Intern v Apple Intern v Gender
59.50%
60.00%
60.50%
61.00%
61.50%
62.00%
62.50%
63.00%
63.50%
64.00%
64.50%
65.00%
20 30 50 70
50.00%
52.00%
54.00%
56.00%
58.00%
60.00%
62.00%
64.00%
Confusion Matrix: • Google has low precision and high recall
• Diverse employee body • Likely to classify everyone to
Google • Facebook has high precision and low
recall • Unlikely to classify other
employees to Facebook • Likely to classify Facebook to other
companies
Feature Importance • Apple
• Experienced employees • More long Gme employees • More skills
• Google • More new employees recently • Master degree • Bilingual
• Facebook • More new bloods in the past 5 years
• Internship experience
Contact info
Result Change on Excluding One Feature Effect of Number of IteraGon on Neural Network Results
F1 Score on Different Algorithms
Top Related