WittyWicky Inc. is a consulting firm that designs brand strategy for a lot of product startups. Their modus operandi is to gain the pulse of competing products and the associated sentiment from social media. Social media has profound impact in capturing the potential customers and thus there are a lot of consulting firms that operate in the digital strategy space. Whether it is to design a marketing campaign or look at the effect of marketing campaigns on user engagement or sentiment, it is a very valuable tool.

Manual assessment of sentiment is very time consuming and automatic sentiment analysis would deliver a lot of value. As a team of data scientists consulting for WittyWicky Inc., you are now responsible for meeting their business outcomes.

Twitter has now become a useful way to build one's business as it helps in giving the brand a voice and a personality. The platform is also a quick, easy and inexpensive way to gain valuable insight from the desired audience. Identifying the sentiments about the product/brand can help the business take better actions.

You have with you evaluated tweets about multiple brands. The evaluators(random audience) were asked if the tweet expressed positive, negative, or no emotion towards a product/brand and labelled accordingly.

This dataset contains around 7k tweet text with the sentiment label.

The file train.csv has 3 columns

1. tweet_id - Unique id for tweets.
2. tweet - Tweet about the brand/product
3. sentiment - 0: Negative, 1: Neutral, 2: Positive, 3: Can't Tell

We will be using ‘weighted’ F1-measure as the evaluation metric for this competition