CONSUMER INTENTION PREDICTION USING
TWITTER
•We aim to analyze the
tweets related to a product and identify the purchase intention in it. In this way we can rank the tweets which have high purchase intention and report the
name of the person who tweeted as potential customer of product.
•We will make a model
by gathering tweets from users who have already expressed intention to buy the
product and see their tweet history and if possible, their web search history
as well. Using this model, we will input potential customers who have tweeted
about the product but have not bought it yet! And based on the training data
the model will estimate a prediction/likelihood of whether the customer will
buy it or not.
•Used
scraper to gather tweets
•Used
the following preprocessing techniques:
–1.
LOWERCASE
–2.
REMOVE PUNC
–3.
STOPWORDS REMOVAL
–4.
COMMON WORD REMOVAL
–5.
RARE WORDS REMOVAL
–6.
SPELLING CORRECTION
–7.
STEMMING
–8.
LEMMATIZATION
•Next,
we made 3 types of document vectors:
–1.
TF
–2.
IDF
–3.
TF-IDF
•We
used the following text analytical models:
1.Support Vector Machine (SVM)
2.Naive Bayes
3.Logistic Regression
4.Decision Tree
5.Neural Network
•We aim to analyze the
tweets related to a product and identify the purchase intention in it. In this way we can rank the tweets which have high purchase intention and report the
name of the person who tweeted as potential customer of product.
•We will make a model
by gathering tweets from users who have already expressed intention to buy the
product and see their tweet history and if possible, their web search history
as well. Using this model, we will input potential customers who have tweeted
about the product but have not bought it yet! And based on the training data
the model will estimate a prediction/likelihood of whether the customer will
buy it or not.
•Used
scraper to gather tweets
•Used
the following preprocessing techniques:
–1.
LOWERCASE
–2.
REMOVE PUNC
–3.
STOPWORDS REMOVAL
–4.
COMMON WORD REMOVAL
–5.
RARE WORDS REMOVAL
–6.
SPELLING CORRECTION
–7.
STEMMING
–8.
LEMMATIZATION
•Next,
we made 3 types of document vectors:
–1.
TF
–2.
IDF
–3.
TF-IDF
•We
used the following text analytical models:
1.Support Vector Machine (SVM)
2.Naive Bayes
3.Logistic Regression
4.Decision Tree
5.Neural Network
thank you for your comment
pls call me on 8125424511