A detection mechanism with text mining cross correlation approach
José Luís, Guerrero Cusumano · 2017
Text mining is a knowledge-intensive task. Text mining practitioner finds many analytics, statistics and linguistics concepts very difficult to deal with and it is compounded by additional difficulties inherent to the user's native language and culture. Correspondence analysis (CA), SVD plot and cross-correlation analysis (CCA) are used to analyze concepts and words, correlated in time and to create a warning/forecast mechanism. We apply CA and CCA to Toyota's quality control problems related sudden acceleration of their cars. The predictive model a “cross correlated autoregressive” created helps to determine the number of lags (weeks) which could have been used for the detection of Toyota Accidents.