“A Predictive Model for Risk Security Manufacturing Plant in Mexico

Industrial Security in manufacturing facilities is a common, relevant matter for most industries, as this factor may affect the inner general environment of the firm, the production rate and its stability, the position of the firm as an attractor of talent as well as a trigger of incremental federal contributions from the firm to the National Public Social Security System (IMSS, Instituto Mexicano del Seguro Social, for its acronym is Spanish). This happens in the case of Mexico, the geographical context in which this research was developed. All these implications redound on strong impact in the profitability of firms.
The purpose of this research focused in developing a predictive model for estimating the occurrence of unsafe acts, unsafe actions and disabling and non-disabling injuries (negative outputs). Supported by a literature review, research team concluded on a set of causal factors of negative outputs during work shifts of a selection of production lines in a manufacturing facility in Mexico. Using existing data for causal factors, and capturing other non-existing using questionnaires, team developed a database of more than 2400 useful records (reporting occurrences of interactions of man, machine and environment of work shifts, on a daily basis, three work shifts per day) during the first week of July, 2018 and the last week of October, same year. Using this data, and implementing analytics predictive techniques such as Logistics Regression, Support Vector Machine and Averaged Perceptron, research team estimated a model for predicting negative outputs analyzing interactions (man, machine and environment) of work shifts. Implications of research findings focus on preventing unsafe acts, unsafe actions and disabling and non-disabling injuries, based on the estimation of probability of negative occurrences in the context of work shifts.