Q 646. Predictive modeling is a commonly used statistical technique to predict future behavior. How far in the future can this technique be used to predict the process behavior? Provide some examples to support your answers.

 

Note for website visitors -

• This platform hosts two weekly questions, one on Tuesday and the other on Friday. 
• All previous questions can be found here: https://www.benchmarksixsigma.com/forum/lean-six-sigma-business-excellence-questions/. 
• To participate in the current question, please visit the forum homepage at https://www.benchmarksixsigma.com/forum/.
• The question will be open until Tuesday or Friday at 5 PM Indian Standard Time, depending on the launch day. 
• Responses will not be visible until they are reviewed, and only non-plagiarised answers with less than 5-10% plagiarism will be approved. 
• If you are unsure about plagiarism, please check your answer using a plagiarism checker tool such as https://smallseotools.com/plagiarism-checker/ before submitting. 
• All correct answers shall be published, and the top-rated answer will be displayed first. The author will receive an honorable mention in our Business Excellence dictionary at https://www.benchmarksixsigma.com/forum/business-excellence-dictionary-glossary/ along with the related term.

• Some people seem to be using AI platforms to find forum answers. This is a risky approach as AI responses are error prone as our questions are application-oriented (they are never straightforward). Have a look at this funny example - https://www.benchmarksixsigma.com/forum/topic/39458-using-ai-to-respond-to-forum-questions/

• We also use an AI content detector at https://crossplag.com/ai-content-detector/. Only answers with less than 15-20% AI-generated content will be approved.

Predictive Modelling:

It is the most prevalent statistical technique used to predict future behaviour.Predictive Modelling also contemplated as a mathematical process used to forecast future outcomes and events by analysing the pattern in a data set.It works by analysing past data,current data and projecting the leanings on the result which has been generated to forecast the outcomes. Predictive Modelling can be taken as assumptions which relies on what is happening currently and what has happened in the past.

Here are the 5 models of Predictive Modelling:

Forecast model: Its a very common model and it works on the values which based on numbers got from the learning of historical data.

Time series model: This works on sequence of points in the data given based on time.

Classification Model: This model merely works on classification of data into different types of categories from the past and current data and analyses the future outcome from it.It is the most simplest model of predictive Modelling.

Clustering Model: This model works on common attributes on the same data.Like grouping the smaller things,people with same behaviour will be considered as a sub group from a large scale data.

Outliers model: This works on outlying data or analysing abnormal data points.This technique can be used to predict the process behaviour as short term predictions, medium term predictions and long term predictions.

Example for short term prediction: prediction of traffic on websites for the next hours so that it helps in utilisation of server resources.

Medium term prediction: Predicting monthly power consumption on a building to optimize HVAC systems.

Long term prediction: Predicting occurrence of natural disasters like earthquakes, hurricanes floods and droughts.

Example for predicting models:

Fraud detection in banks, insurance companies:

Predictive Modelling helps to detect fraudulent activities,transactions like insurance or credit card fraud 

Predictive Modelling is a widely used technique in the field of AI/ML by organizations to predict the future outcome of a process/ service across industries. It predominantly uses historical data, identifies patterns and uses Machine learning models like Ensemble Techniques to predict a future outcome. The output can be further analyzed to proactively take measures to avoid an impact. Some of the examples where predictive modelling is successfully implemented are

1.       Banking – To identify defaulters to repay loans

2.       Telecom – In the Network monitoring area to identify failure of active/ passive infrastructure

3.       Healthcare – Identify the critical illnesses in the patients

4.       IoT/ OTT Platform – Proactively suggesting content-based user preferences and usage

5.       Agriculture – Predicting the soil fertility, crop growth harvesting and weather patterns

Here are my thoughts on how Predictive modelling can be used to predict process behavior.

A process is a step-by-step procedure or a set of activities that need to be executed either in sequence or parallel that results in the delivery of a service or the creation of a product. The Process involves three critical components People, Procedures and Platforms.

People – The resources who are involved in delivering a service or the creation of a product

Procedures – The documentation the people use to deliver a service or a creation of a product

Platforms – The tools and applications that are involved in delivering a service or a creation of a product

All the above three involve and generate a large amount of data which is a vital part of Predictive Modelling or Predictive Analytics. Here are some examples of how Predictive Modelling can help predict the process behavior

People: This component has a great influence on delivering the process successfully, Predictive Modelling can be implemented in the following areas to predict the Process Behavior in any BPO/ KPO

1.       Process Quality – Based on patterns we can predict when a process can have quality issues in the production like when there is an increase in complex orders, changes in the resource skillset

2.       Attrition – Based on skillset and the demand in the market we can predict the attrition patterns which could impact process failure

3.       Knowledge Retention – Based on volume patterns we can predict the need to cross-functionally train resources for effective utilization, for a short period we can’t onboard resources where cross-functionally trained resources can meet the demand

Procedures: is very vital to deliver a Process on time with quality. Here are some examples where a Process Behavior can be predicted using Predictive Modelling

1.       Failures – Using historical data like volume and quality we can predict when a process could potentially fail due to the lack of procedures like un-updated SOPs

2.       Change – When a process could require a change like the need for process re-engineering to revise a benchmark and update procedures

Platforms: is technology driven and would be constantly upgrading as the technology evolves. To study the Process Behavior Predictive Modelling can be implemented in the following areas

1.       Uptime – The Customers are constantly expecting the services to be available round the clock, Predictive modelling can help to predict how platform-related failures can be eliminated that could impact a Process

2.       Consolidation – When the platforms can be consolidated to achieve process efficiency

Predictive modeling is a statistical technique to predict future behavior. Its solutions are a type of data mining technology. It analyzes historical and current data and generates a model to predict future outcomes.

It is used to predict customer behaviors, Financial, economic and market risks.

There are different Types of Predictive modelling:

·       Classification Models: This type of model uses machine learning and places data into categories. These are of several types, few of them are:

o   Logistic regression - Yes/No estimation of future events

o   Decision Trees – in addition to yes/No decision, it deals with if/else to predict future events.

o   Random forest – it combines unrelated decision trees using classification and regression.

o   Neural networks – Large volume of data is reviewed to reveal correlation.

o   Naïve Bayes: it is based on Bayes' Theorem, which forms conditional probability.

 

·       Clustering Models: In this model, the data points are grouped. According to this model, the data in one group should have similar characteristics when compared to the data in different groups. Below are few of the clustering models-

o   K-Means: This model identifies central tendencies of different groups of data.

o   Mean-Shift: In this model, the mean of a group is shifted.

 

·       Outlier Models: The data always has outliers. The outliers are identified by the following few algorithms:

o   Isolation Forest

o   Minimum Covariance Determinant (MCD.

o   Local Outlier Factor (LOF)

·       Time Series Models: It uses historical data to predict future events Below are few of these models:

o   ARIMA: uses autoregression integration and moving averages to predict results.

o   Moving Average uses an average of a specified period.

Advantages Of Predictive Modelling

• It is an easy way to generate actionable perceptions.
• It is used to test different scenarios.
• It increases the speed of decision-making.

Disadvantages Of Predictive Modelling

• Calculations cannot be explained.
• Bias due to human input
• High learning curve

Predictive modeling is a commonly used mathematical or statistical tool or technique used to predict future behavior. Predictive modeling uses historical data and analytic techniques along with machine learning to generate a model which helps in predicting the future outcomes. In predictive modeling, data is collected, a statistical model is formulated, predictions are made, and the model is validated (or revised) as additional data becomes available.

 

With the help of predictive analytic tools and model formulated, any organisation can predict the outcome, by using past and current data, in days, month or even years in future. 

 

Also, it helps in identifying patterns hidden in data and simplyfy the same into risks and oppurtunities lying ahead in future. 

 

Examples of Predictive Modelling:

Aerospace: Predictive modelling helps in analysis of effect of maintenance activity on operations, fuel consumption pattern, etc.

Automotive: Predictive modelling helps in analysis of records of component robustness, failure cases, malfunctioning in parts and incorporate those learning in upcoming manufacturing designs and plans. It also helps in studing driver behavior to create better diver assist system or smart vehicles.

Energy: It helps in forcasting the consumption pattern of different energies, demand and supply pattern, etc.

Financial services: It helps in predicting market trends, development of credit risk model, determining the spend pattern of demography, etc.

Manufacturing: It helps in predicting the maintenance schedule, helps in strengthening the supply chains by prediction of demand supply graph, etc.

 

 

 

 

 

Most of the respondents have answered what is Predictive Modelling, however except for a couple of answers, none of the others have answered the question on how far in the future can this technique be used to predict the process behavior. With the advancements in data analytics, predictive modelling can now be used to predict with reasonable confidence for short and medium terms. Longer term predictions are prone to high inaccuracies and lower confidence. The other thing worth noting is that it is a continuous learning activity. As we gather more historical data, as the business dynamics change, the model has to be retrained so that high confidence predictions can be made.

 

The closest answer is given by Anvitha and hence has been selected as the winner.