A new Twitter based credit rating model methodology.

Second, we select features with a low p-value because this means that these linguistic features are highly significant which is a promising attribute in terms of predictive accuracy.

also in 132,142 other papers