Description
Stack Overflow is the largest, most trusted online community for developers to learn, share their programming knowledge, and build their careers.
Stack Overflow is something which every programmer use one way or another. Each month, over 50 million developers come to Stack Overflow to learn, share their knowledge, and build their careers. It features questions and answers on a wide range of topics in computer programming. The website serves as a platform for users to ask and answer questions, and, through membership and active participation, to vote questions and answers up or down and edit questions and answers in a fashion similar to a wiki or Digg. As of April 2014 Stack Overflow has over 4,000,000 registered users, and it exceeded 10,000,000 questions in late August 2015. Based on the type of tags assigned to questions, the top eight most discussed topics on the site are: Java, JavaScript, C#, PHP, Android, jQuery, Python and HTML.
Problem Statemtent
Suggest the tags based on the content that was there in the question posted on Stackoverflow.Source: https://www.kaggle.com/c/facebook-recruiting-iii-keyword-extraction/
Data Source : https://www.kaggle.com/c/facebook-recruiting-iii-keyword-extraction/data
Youtube : https://youtu.be/nNDqbUhtIRg
Research paper : https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/tagging-1.pdf
Research paper : https://dl.acm.org/citation.cfm?id=2660970&dl=ACM&coll=DL
- Predict as many tags as possible with high precision and recall.
- Incorrect tags could impact customer experience on StackOverflow.
- No strict latency constraints.
Refer: https://www.kaggle.com/c/facebook-recruiting-iii-keyword-extraction/data
All of the data is in 2 files: Train and Test.
Train.csv contains 4 columns: Id,Title,Body,Tags.
Test.csv contains the same columns but without the Tags, which you are to predict.
Size of Train.csv - 6.75GB
Size of Test.csv - 2GB
Number of rows in Train.csv = 6034195
The questions are randomized and contains a mix of verbose text sites as well as sites related to math and programming. The number of questions from each site may vary, and no filtering has been performed on the questions (such as closed questions).
Title: Implementing Boundary Value Analysis of Software Testing in a C++ program? Body :\n\n#include< iostream>\n #include< stdlib.h>\n\n using namespace std;\n\n int main()\n {\n int n,a[n],x,c,u[n],m[n],e[n][4];\n cout<<"Enter the number of variables";\n cin>>n;\n\n cout<<"Enter the Lower, and Upper Limits of the variables";\n for(int y=1; y<n+1; y++)\n {\n cin>>m[y];\n cin>>u[y];\n }\n for(x=1; x<n+1; x++)\n {\n a[x] = (m[x] + u[x])/2;\n }\n c=(n*4)-4;\n for(int a1=1; a1<n+1; a1++)\n {\n\n e[a1][0] = m[a1];\n e[a1][1] = m[a1]+1;\n e[a1][2] = u[a1]-1;\n e[a1][3] = u[a1];\n }\n for(int i=1; i<n+1; i++)\n {\n for(int l=1; l<=i; l++)\n {\n if(l!=1)\n {\n cout<<a[l]<<"\\t";\n }\n }\n for(int j=0; j<4; j++)\n {\n cout<<e[i][j];\n for(int k=0; k<n-(i+1); k++)\n {\n cout<<a[k]<<"\\t";\n }\n cout<<"\\n";\n }\n } \n\n system("PAUSE");\n return 0; \n }\n<p>The answer should come in the form of a table like</p>\n\n <pre><code> 1 50 50\n 2 50 50\n 99 50 50\n 100 50 50\n 50 1 50\n 50 2 50\n 50 99 50\n 50 100 50\n 50 50 1\n 50 50 2\n 50 50 99\n 50 50 100\n </code></pre>\n\n <p>if the no of inputs is 3 and their ranges are\n 1,100\n 1,100\n 1,100\n (could be varied too)</p>\n\n <p>The output is not coming,can anyone correct the code or tell me what\'s wrong?</p>\n'Tags : 'c++ c'
It is a multi-label classification problem
Multi-label Classification: Multilabel classification assigns to each sample a set of target labels. This can be thought as predicting properties of a data-point that are not mutually exclusive, such as topics that are relevant for a document. A question on Stackoverflow might be about any of C, Pointers, FileIO and/or memory-management at the same time or none of these.
__Credit__: http://scikit-learn.org/stable/modules/multiclass.html
Accuracy is used when the True Positives and True negatives are more important while F1-score is used when the False Negatives and False Positives are crucial. In most real-life classification problems, imbalanced class distribution exists and thus F1-score is a better metric to evaluate our model
Micro-Averaged F1-Score (Mean F Score) : whenever we want to optimize both precision and recall one good metric to do so is F1 score.The F1 score can be interpreted as a weighted average of the precision and recall, where an F1 score reaches its best value at 1 and worst score at 0. The relative contribution of precision and recall to the F1 score are equal. The formula for the F1 score is:
F1 = 2 * (precision * recall) / (precision + recall)
But normal f1 score is valid only for Binary class classification problems.
In the multi-class and multi-label case, this is the weighted average of the F1 score of each class.
'Micro f1 score':
Calculate metrics globally by counting the total true positives, false negatives and false positives. This is a better metric when we have class imbalance.
'Macro f1 score':
Calculate metrics for each label, and find their unweighted mean. This does not take label imbalance into account.
https://www.kaggle.com/enforcer007/what-is-micro-averaged-f1-score
http://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html
Hamming loss : The Hamming loss is the fraction of labels that are incorrectly predicted.
https://www.kaggle.com/wiki/HammingLoss