Description
IS420UCSCH2A GE Mark VI
As the functionality of distributed computing tools such as Spark MLLib (http://spark.apache.org/mllib) and SparkR (http://spark.apache
.org/docs/latest/index.html) increases, it becomes It is easier to implement distributed and online machine learning models, such as support
vector machines, gradient boosting trees and decision trees for large amounts of data. Test the impact of different machine parameters and process
measurements on overall product quality, from correlation analysis to analysis of variance and chi-square hypothesis testing to help determine the impact of individual
measurements on product quality. This design trains some classification and regression
models that can distinguish parts that pass quality control from parts that do not. The trained models can be used to infer decision rules. According to the highest purity rule,
purity is defined as Nb/N, where N is the number of products that satisfy the rule and Nb is the total number of defective or bad parts that satisfy the rule.
Although these models can identify linear and nonlinear relationships between variables, they do not represent causal relationships. Causality is critical to
determining the true root cause, using Bayesian causal models to infer causality across all data.
3.3 Visualization
A visualization platform for collecting big data is crucial. The main challenge faced by engineers is not having a clear and comprehensive overview of the complete manufacturing
process. Such an overview will help them make decisions and assess their status before any adverse events occur. Descriptive analytics uses tools such as
Tableau (www.tableau.com) and Microsoft BI (https://powerbi.microsoft.com/en-us) to help achieve this. Descriptive analysis includes many views such as
histograms, bivariate plots, and correlation plots. In addition to visual statistical descriptions,
a clear visual interface should be provided for all predictive models. All measurements affecting specific quality parameters can be visualized and the data
on the backend can be filtered by time.
Reviews
There are no reviews yet.