Deep Dave

Parametric Analysis is statistical analysis done on known distributions, where we can make inference on population parameters like mean and standard deviation based on sample statistic. Exp. Normal Distribution.

There are some key features of Parametric Analysis:
 
1. Assumption: It follows specific know distribution.
 
2. Efficiency: If efficiency holds true, it provides precise results.
 
3. Rely on Statistical theories and formulas.
 
Usage Across Industries:
 
1. Pharmaceuticals & Healthcare Industries: In pharma & healthcare for new drug development we need clinical trial and bio equivalence study which heavily relies on parametric analysis as it requires inference based on sample data provided that the data follows known data distribution.
 
2. Manufacturing & Quality Control: For developing robust product we need to ensure that process parameters have high sigma level. Generally, we try to increase sigma level of Critical
Quality Attributes (CQAs) through optimising Critical Process Parameters (CPPs).
 
3. Aerospace & Automotive: High reliability is a must requirement for aerospace & automative industries which requires high precision and accuracy which means it has high application of parametric analysis.
 
4. Service Industries: In service industry, the arrival rate, service rate and queuing follows specific statistical distributions that also requires parametric analysis.
 
Advantages of Parametric Analysis:
 
1. High Precision: In manufacturing, if we  want to compare machine’s output to specific standard the parametric t-test is the best option.
 
2. Power of Test: In pharma, while conducting clinical trials the parametric tests like ANOVA can help with checking efficiency of new drug.
 
3. Wide Applications: Most real life applications follow normal
distribution which is helpful for regression modelling (Exp- Energy Prediction Modeling)
 
4. Simple Interpretation:  Process capability analysis through Cp & Cpk helps with summarising how the process is performing against customer
defined specification limits.
Exp- Moisture content in drug can be measured in Cp & Cpk and can be interpreted on how it performs against specification limits.

In real life scenarios we witness many such failures in a Tech Product or Tech Tool where root cause can be linked to a flaw in the Business Requirement Document (BRD). BRD is a document that articulates product or tool requirement to solve business problem or improve business performance and this document is taken as a guidebook by supplier to develop Tech product or Tech tool.
 
BRD generally consists of the feature, functionality, performance expectation of the product and also the constraints of the project. Thus, it's clear that any ambiguity or inaccuracy may result in potential failure in product or tool. There can be multiple reasons for this:
 
1. Lack of Articulation: Articulating the exact requirement in BRD is a tough task. Generally, the user has dream Tech product conceptualization in mind but articulating that in exactly the same way can play significant role in success or failure of Tech product or Tech tool. 
Example: Apple's BRD on Maps could not convey the requirement of precise location with accuracy that resulted in significant map errors, improper navigations and low quality satellite images.
 
2. Lack of Prioritization: In many cases people want it all in BRD and give all the possible features in Tech product or Tech tool which may result in lack of prioritization to unique and most valuable features.
Example: Windows Vista is a perfect example for this where BRD listed so many features without clear prioritization on performance focus. Ultimately, it failed due to frequent crashes and compatibility problems and had to be replaced with Windows 7.
 
3. Misinterpretation of Business Needs: In many BRDs specificity is missing which may create misinterpretation. 
Example: If requirement is on support with multiple payment options for e-business application, then developer might miss one of the important integrations (exp-Bit Coin payment option) which may result in loss of potential sales.
 
4. Poor Definition of Success Criteria: This, I believe is the most important reason as Success Criteria for the Tech product or Tool must be defined with respect to performance, quality, durability and cost effectiveness. 
Example: JioChat where despite so many Jio Users the app could not succussed as it was lacking success criteria like target for user adoption, engagement and market spread.
 
Thus, it is important to articulate 360deg angle on Business Requirement in BRD so there is zero misalignment.

No matter how advanced today's Large Language Models (LLMs) are, there are many limitations due to which LLMs are not ideal to solve all kind of problems. Let's review some of the problem categories with examples which LLMs cannot solve:
 
1. Problems Requiring Reasoning & Inference: LLMs depend on matching the pattern and lack the deep understanding like human brain. So, in many problems where in-depth understanding and logical reasoning are required, LLMs fail to solve the problem.
 
Example: Apple did one study, where it demonstrated failure of LLMs in standard mathematical problem where in by changing names and numbers led to drop in performance up to 9.2% and when irrelevant details were included, there was drastic drop to 65.7%.
 
2. Problems Requiring Maintaining Context in Long Conversations: Many a times, we need to give a long context to LLMs for getting a response considering all the constrains and entire context. But they most of the time loose the context in a long and extended conversations.
 
Example: In ChatGPT (If you are extensively using it, then you know!) when we give long context, many a times it forgets earlier context then we again need to reiterate to consider that and reshare the response.
 
3. Problems With Ambiguity and Incomplete Information: If there is incomplete or ambiguous inputs are given then LLMs may generate incorrect or nonsensical responses.
 
Example: If you are typing open ended question like "What to do if my car is not starting?", the LLM response might suggest checking the battery, but it may also suggest unrelated actions like checking the oil due to no situational understanding. Here, Situational Understanding becomes important as it eliminates many probable root causes in the problem.
 
4. Problems Requiring UpToDate Knowledge: LLMs are generally good at solving problems using historical data & context till which model is trained, but there is no real time updating capabilities. This means LLMs might give some outdated information.
 
Example: LLM trained before 2024 will not be aware of latest events, leading to inaccurate responses.
 
5. Biased or False Information: LLMs has large backend data upon which it is developed. If biased data is used in the backend, then LLMs will generate biased or false information which is dangerous.
 
Example: This is recently in news that the countries developing LLMs feed the back-end data as per country's stand on controversial topics (e.g. Land, Wars etc.). Two countries may have different LLMs giving different response to same question as who won the historical war between A & B?
 
 
 
 
 
 
 

Retrieval-Augmented Generation (RAG) and fine-tuning for LLM-Powered Agent (Large Language Models) are both types of AI language generation methods of Natural Language Processing (NLP) with some fundamental difference in response generation. Let's understand:
 
In RAG, the response generation is based on external source of knowledge that helps with real-time information update & the same information is used to augment the model's response. Basically, instead of relying on internal knowledge used for model training, RAG has kind of "Open-Book" approach where any additional or real-time information is looked from external sources.
 
Whereas in Fine-tuning for an LLM-Powered Agent, there is pre-existing internal data source which is fine-tuned through continuous improvement on a specific set of data to specialize it for specific task or domain. Say for example, we are adding specific books in internal data sources for generation of more advanced & niche responses which industry or domain specific.
 
Application for RAG:
Dynamic & real-time knowledge requirements (e.g. stock price, latest news, latest research etc.) Model Size Reduction (as no need for huge internal data source required) Cost & Efficiency (requirement of low or no internal data eliminates requirement of high energy consuming servers for storing internal database) Application for Fine-Tuning with LLM Powered Agent:
Nuances & Specialized Tasks (e.g. application with special domain language, knowledge and jargons like medical, legal etc.) Consistent Tone and Style (nuanced in line with ask in the prompt) Hybrid Use Case (RAG & Fine-Tuning with LLM Powered-Agent)
 
In cases where we need the response that is specialized in language and also dynamic with respect to real-time or latest information update, we need both internal and external sources for response generation. In this scenario, we need to explore the Hybrid model.
 
Examples: Let's say we have created Financial Advisor Chatbot which uses finance related language and tax laws using (Fine-Tuning with LLM) as well as gives advice considering latest market dynamics (stocks, bonds, investment avenues etc. - RAG). In this case, it's best to use both internal and external sources to generate ideal response.
 
To summarize, hybrid model is ideal scenario for specialized and real-time response generation.
 
 

Replication followed by benchmarking is one of the most effective and the easiest methods for lean six sigma project identification. The greatest advantage here is that we get to replicate the projects with proven success and results.
 
However, if we seek replication of patented products or processes then we may face legal challenge due to violation or infringement of laws related to Intellectual Property Rights (IPR).
 
Let's see below the major challenges and strategies or approaches to overcome these challenges:
 
Major Challenges in Replication of Patented Products (DFSS) or Processes (Six Sigma):
 
1. The Legal Aspect: The IPR laws limit us to replicate the design or technology used behind a patented product or process. Failing to comply this our company may face lawsuits and loss of reputation.
 
2. The Innovation Aspect: Companies that heavily rely on replication may miss the huge opportunities of inhouse innovation. Different companies have different goals and strategies, in these scenarios replicating technology or process patents dealing with specific issues or goal of other company may not be relevant to us. Instead, the companies should train people and start an innovation drive of their own.
 
3.The Information Aspect: Generally, the detailed information on "How?" part is missing behind patented products or processes. In this case replicating the exact concept is a challenge.
 
Strategies & Approaches to Overcome Challenges on Replication of Patented Products (DFSS) or Processes (Six Sigma):
 
1. Adaptation of Principles Instead of Process or Technology: Instead of replicating exact process or product design, can we think of replicating principles behind? This helps with avoiding legal risk as well as capturing the essence of patent technology.
 
2. Collaborate with Patent Filing Organization: In the cases where it is difficult to replicate, we can think of collaborating with patent filing organization through licensing agreement. This is the best approach when we do not have internal capability to replicate the exact patent technology.
 
3. There are Multiple Solutions to a Single Problem: While we know this fact, we can train & encourage inhouse team to innovate different solutions to the same problem.
 
4. Innovation In Six Sigma: In Six Sigma Projects, typically in Improve phase when we explore different improvement initiatives, bringing in innovative thinking and methodologies (exp. SCAMPER) can help the teams identify innovative solutions. This way we can avoid necessity to replication to patent technology and also get the end result.
 
To summarize, we have strategies like collaboration, adaptation, innovation, inhouse R&D and capabilities development to deal with the challenges.