hunsnowboarder Goto Github PK

ba-pre-session-2019

Pre-session for Business Analytics 2019

company_growth_predictor

# Company_growth_predictor task was to predict which companies are growing fast and which not -> classifier model

corrupted_document_challenge

Problem context An English document has been corrupted in an unusual manner. Each alphabetic word (no numbers, but apostrophes allowed) has had every other character replaced with a # character. For example, a sentence like I'm here. It's a cat. would appear in the document as I#m h#r#. I#'# a c#t. We have extracted the tokens from this corrupted document and have provided the list of tokens to you as corrupted_tokens.txt. Note that the order of tokens in the list is the same as how they appeared in the original document, and that surrounding punctuation like . and , have been removed. As an example, if the uncorrupted document is: I'm here. It's a cat. the its corresponding corruption followed by tokenization would be: I#m h#r# I#'# a c#t For further details please read the accompanying pdf file.

fashion_mnist_challenge

Solving the well-knonw fashion mnist dataset challenge with CNN

gdsa

Grapevine assignment

hot_dog_not_hot_dog

nlp_article_classification

# NLP_project The goal of the project is to classify an article based on its different features if is popular or not. The articles were extracted from the Mashable main site. For more information on the analysis and outcome please check the accompanying python notebook file. The final model (gradient boosting) achieved a 0.7168 roc auc in predicting if an article will be popular or not (results cross validated on test dataset).

sentiment_analyis_aws_cloud

Sentiment analysis on Obama's and Trump's speech on killing of Bin Laden and al-Baghdadi

turnover_prediction

A very basic turnover prediction algorithm.