GitHub - autonise/Small-snipet-classification: Classification of small snipets using facebook fast-text,

#Topics Explored by us

#Results

The results have been tabulated in the File Results for all the grid runs.

FastText gave the best results
Naive Bayes approach gave intermediatery results
Convolutional Neural Networks performed the worst because of overfitting
LDA performed well for data that is well represented in the data
Guided LDA topics are no better than LDA
40 optimum clusters are idetified for clustering approach which led to 20 topics
Identified topics are better than Guided LDA and match the LDA approach

Name		Name	Last commit message	Last commit date
Latest commit History 16 Commits
.ipynb_checkpoints		.ipynb_checkpoints
LDA		LDA
.DS_Store		.DS_Store
Clustering Unsupervised.ipynb		Clustering Unsupervised.ipynb
Convolutional Neural Networks.ipynb		Convolutional Neural Networks.ipynb
Fast Text.ipynb		Fast Text.ipynb
Naive Bayes.ipynb		Naive Bayes.ipynb
README.md		README.md
Word2Vec.ipynb		Word2Vec.ipynb
data.txt		data.txt
final_data_google.p		final_data_google.p
grid_results.txt		grid_results.txt
model.bin		model.bin

Provide feedback