[x] Clean Twitter API response based on the type of data returned. (may need to remove @, urls etc.)
[x] Decide if there's a need to make a separate model from when the user enters text on the browser. This may be needed coz the cleaning function, the predict function will all be different.
[x] Send the twitter response as a DataFrame to the model. Even a Pandas series should do.
[x] Batch predict the top n tweets and run a majority vote on the predictions. (This won't work while fetching probabilities so we will need to probably normalise the sum of probabilities of each type returned)
[x] Decide n by validation.
[x] Check if expected response is being returned by the server.
[x] Clean Twitter API response based on the type of data returned. (may need to remove @, urls etc.)
[x] Decide if there's a need to make a separate model from when the user enters text on the browser. This may be needed coz the cleaning function, the predict function will all be different.
[x] Send the twitter response as a DataFrame to the model. Even a Pandas series should do.
[x] Batch predict the top
n
tweets and run a majority vote on the predictions. (This won't work while fetching probabilities so we will need to probably normalise the sum of probabilities of each type returned)[x] Decide n by validation.
[x] Check if expected response is being returned by the server.