Testing the efficiency of Sentiment Analysis models

I used the sentiment analysis model here to analyse the [amazon reviews dataset](https://www.kaggle.com/bittlingmayer/amazonreviews).

However, some of the data leads to a BoundsError
```
ERROR: BoundsError: attempt to access 32×5000 Array{Float32,2} at index [Base.Slice(Base.OneTo(32)), 5001]
```
For those, I skipped and ignored them.

Hence, out of the 400000 sets of data, 399703 were interpreted, and out of those, 207103 were interpreted correctly by the model, hence giving an accuracy of 51.814222059879455%

Accuracy: 0.5181422206
Precision: 0.5159145929
Recall: 0.5916847176
F1 Score: 0.5512079636
(Thanks PseudoCodeNerd for introducing these numbers to me! Find more about them [here](https://towardsdatascience.com/the-5-classification-evaluation-metrics-you-must-know-aa97784ff226))

Thanks :)

Related to #160 #184 #84 #182