Sanat Kumar

A3 which is commonly referred in terms of paper size “11.7*16.5” inches. A3 is also an important problem solving “lean management” tool introduced by Toyota. A3 Report is a single page document, which highlights the problem and solution for whole process. There are 7 Major elements of A3 Report are: 1) Background – highlighting the problems at hand 2) Current Situation – elaborates and highlights current baseline performance. Also highlight “As-is situation” 3) Target Setting – Establishing and validating target the improvement opportunity 4) RCA- Root cause analysis of the problem identified using different technique 5) Improvement plans- Project team based upon the RCA define the improvement plans 6) Implementation – Creating implementation plan to deploy improvements 7) Follow up- Conduct training, points to share, pending actions so on Apart from that other component “Team Charters”, “Scoping”, “CTQ”, “Defect Definition” could be added as part of A3 solution. How does A3 format helps? As per the famous quote by Abraham Lincoln, “Give me six hours to chop down a tree and I will spend the first four sharpening the axe.” Same is the purpose of A3, instead of looking problem in isolation, A3 problem solving technique advices that process should sit together and try to understand the problem at hand, do a deep dive analysis to find the root cause (which requires collaboration of the team which could be cross functional) before designing the improvement plans. Benefits of A3 problem technique · Promotes the culture of reaching the root cause of the problem, rather than working symptoms · A3 report helps in presenting the case to higher management, used for training and coaching · Promotes collaborative effort by team · Solution deployed are discussed with the project team hence reduces the risk · Structured method and ease to use · It is easily modified and could be replicated

Feature Creep in general means adding more feature to existing features, which might add complexity to the product use. In general every product has a product life-cycle, where post maturity it tends to follow decline. But to extend the maturity phase normally new features are added. Eg- Smart Phones where on a regular basis new features are added to ensure that maturity phase could be extended At times addition of multiple features in the product which make is non-user friendly and complex which is termed as “Feature Creep”. The focus of the product owner should be on Quality and user friendliness of the product rather than Quantity. Disadvantages of Features Creep: 1) Its misuse or waste of money and resources of the organization (as time is spend on development, testing, launch so on) 2) Makes product complex and difficult to use 3) If features are not properly thought through consumers will not appreciate it, and might lead to dis-satisfaction rather delighter 4) At times increased features might boost initial sales but later customer lead to customer dis-satisfaction 5) Non user friendly Reason for Features Creep: 1) Competitors have the feature which the product might not have, or competitor might simplified the features 2) Customer requirement might not have been clear with development team 3) Pressure from competition to add more and more features At times Organization or Product owner should understand when to put a halt on adding more features and focus on Quality of the product. Steps which R&D team should take to reduce/eliminate Feature Creep are: 1) Clear understanding of customer requirements (which can be collected via customer feedback, or primary research) 2) Create a strategy for the product- stay focused invested on the key priorities 3) Invest time and resources on to key priorities 4) Frequently analyze customer feedback 5) Focus more on Future gains then short term gains 6) Analyze competitors product (or market leader- if the Organization is not market leader) 7) Do rigorous testing of Alpha and Beta testing 8) Be proactive in analyzing un-necessary feature and scrap it

TRIZ – is theory of inventive problem solving. This was developed by Russian Scientist Genrich S. Altshuller post his extensive study of inventions. He derived 39 conflicting features (matrix) and 40 inventive principles to resolve/or provide direction to solution. Based on the problem team is asked to select the conflicting features and those features highlight the principles (which could be multiple), team then require brainstorming on the solutions using the principle. Smart Little People is tool devised in TRIZ which uses a pictorial idea generation technique. This technique uses miniature dwarf (little smart people) to describe the conflicting situations, which helps people to come out of their stereotype approach, habits. People start looking at the problem at hand in completely different way For Eg- If the problem is “Food grain prices are going high and also the current transportation laws are regulated in terms is using heavy trucks (means limited quantity can only be transported)”. Company wants to optimize the cost. In traditional method, employee might think that this is something which is out of scope (regulations/laws). Where as in Smart Little People approach it will be different where the problem will be generalized to reduce employee biasness. Instead of food grains it will be pebbles and instead of trucks be boxes. Now the contradiction become transportation of pebbles in boxes Contradiction 1- Increase the box size to move more pebbles, bit it might be difficult to move due to weight Contradiction 2- Small box sizes means less pebbles could be moved Now due to this the current problem has been simplified and also the employee will look at the current problem with any pre-conceived idea. As next step team will brainstorm and come out with creative ideas to move the pebbles in the boxes, without thinking of the actual problem. This creative ways helps in: 1) Breaking stereotypes and pre-conceived notions 2) Presenting problem in simplified way helps people to come out with innovative ideas 3) It saves time against the traditional method to trail and error

Business Process - set of sequential steps followed to obtain a desired result. Whereas Business Process Re-engineering - steps taken to de-design the current process in order to meet Organization’s goal and objective (cost, quality, service, speed so on). This concept is adopted in early 1990. BPR is a 4 step methodology: Step 1 is to identify process One of the toughest questions is to identify the process. There are several processes in an organization which can be categorized as “Core and Support process”. BPR concepts should be applied on Core process for the better results. Core processes are those which are related to vision of the organization. Core process (es) provides organization competitive advantage, creates value for customer and are aligned to organization vision Eg- In Banks core process is providing financial services. For a food processing company any process related to production, packaging and shipment of food are core processes. For a back office completing the transaction on time and with minimal error are the core processes. All the other process could be termed as Support process Step 2 Review and Analyze In step 2 “As-is” core processes is reviewed and if required perform some process improvement and see if the output is improving. Even after the changes if the output (desired result) is not impacted much analyze the reason for the low output. Use tool like 5Why analysis to analyze the root cause of the As is processs Step 3-Design To be As part of this step “Design To be process” which might require changes in People/Processes/Technologies. Tools used here to design new process as “Benchmarking” the process which best in class within same industry or different industries Step 4 -Test and Implement Once “To be” process is defined, next step is pilot testing (the new concept is tested on a very small area or in controlled environment). Once the pilot testing is successful, implement the “To-be process in one go (if process identified is small) or in phased manner (if the process is lengthy). Post successful implementation start from Step 1 for the new process re-engineering. In case of failure of pilot test (means that either analysis or the “To-be” process is not right, hence it is re-commended to start from the beginning) DMADV-(Define, Measure, Analyze, Design and Validate) Since BPR focuses on re-think and re-design the process hence traditional Six Sigma DMADV would be on line with the concept. Integration DMADV will enhance the BPR study as Six Sigma focus on the variation reduction and is a customer/business centric approach. Once the core process is identified in Step 1 of BPR its advantageous to introduce DMADV which will help in: 1) Introduction of DMADV will help business to view the process end-to-end from business and customer value point. Hence focus will be more than cost reduction 2) DMADV is a structure approach and is a robust methodology, which also emphasize on cross –functional groups and better stake holder management 3) DMADV will be more focused towards the area where the solution are unknown 4) It’s a metric driven approach hence post improvement using Six Sigma concepts will help in analyzing the actual benefit in terms of metric movement 5) Apart from the as a practice in Six Sigma – it also focuses on sustenance of the improvement Hence DMADV will complement BPR and will result in better yield to the vision of the Organization.

JIT (Just in Time) is one of the Lean Tenet, focuses on reduction or minimal inventories and no idle inventory (inventories could be - raw materials, WIP and finished goods), and hence focuses on waste reduction. To ensure JIT works with true potential there are certain pre-requisite: Sourcing point of view: Supplier/Vendors should also follow JIT and have a highly reliable relationship. Eg.- Supplier should be ready with raw material when asked by the company Supplier should always abide by the lead time agreed and also maintain the quality of the raw material Information should flow from market to supplier (last level), so that even the supplier are aware of the upcoming demands (best case scenario is real time flow of information) Concept of pull should follow then push Transportation and Warehousing: Focus to be laid on the movement of the good rather than warehousing Kanban concept to be used so that during the transportation of good, there is no idle time Raw material should be delivered as close as to the production floor or production area Production: Concept of SMED should be introduced so that wait time could be reduced Product should not have lot of variability or else inventory pile up will be an issue Continuous improvement should be a culture to reduce the processing time People: JIT should be a top down approach Train employee on Lean concepts (especially on JIT) Employee should be cross training Water Spider concept will enable implementation of JIT

Model cross validation is the method commonly used in Machine Learning which helps in estimating the variability (consistency) and reliability (performance of model over a period of time) of a model. While creating a model there are 2 things: 1) Train the data (which means estimating the parameters) 2) Test the algorithm (evaluate the performance) Out of the total data collected partial data is used for creating the model and partial for testing the model. Mostly multiple sets of data are from the population and testing is performed on these sets using different methods which are called as cross training. This helps in understanding which model is reliable and accurate There are mainly two types of cross validation “Exhaustive and Non Exhaustive” Exhaustive – method which tests all possible ways of dividing a sample into various sets used for training and testing 1) Leave p out cross-validation: in this concept p-observations of the data are left and remaining data is used for training the model. This is repeated throughout the original sample Pros: Simple and easy to implement Cons: Time taking approach 2) Leave one out cross-validation: : in this concept 1 observations of the data are left and remaining data”n-1” is used for training the model. This is repeated throughout the original sample Pros: Simple and easy to implement Cons: Time taking approach Non- Exhaustive – method which does not all possible ways of dividing a sample for training and testing 1) K fold cross validation – in this original datasets is divided in K sets and one set is used for testing and “k-1” used for training Pros: Low brassiness, less complexity and entire data is used for training and testing Cons: Not advised for imbalance data sheet 2) Hold out cross validation- Randomly 2 data set is created from Original Data for test and training Pros: Simple and easy to implement Cons: Lot of data is not used for training (creating the model) which might negatively impact accuracy of the model 3) Stratified k-fold cross validation: This concept is used for imbalanced data sheet. Original data is divided in K sets “ensuring one particular class or instance is not over presented when data set is imbalance”. Apart from that set all the other data is used for training Pros: It is used for imbalance data Cons: Not used for time series data

Water Spider or Mizusumashi is instrumental for implementing Lean and removal of wastage Water spiders in lean referred to those individuals who are at a constant move on the job floor to ensure that all the workstations are working smoothly (like the water beetle or water spider, which continuously moves below the surface of the water). He ensures that’s all that the requirements of each work stations are fulfilled and there is continuous flow of production. At times his role could be confused from the Material Management or “go fetch” person but its much bigger than that, his prime role includes: Keeping the workstation stocked with Materials Water spider have the knowledge on the functioning of different work stations “tools and machines” used, and they help in keeping the in-efficiencies of the process at lowest level They are on continuous move between workstations ensuring the flow of production is not impacted (there visit to each workstation is directly proportional to the cycle time of the product) They ensure the movement of Kanban card Even the help in deciding the work load leveling (allocation of workers at work stations) Manage communications They are critical as they eliminate waste like – Transportation – all the material are moved from warehouse to the work stations by the water spider hence transportation as a waste is removed for the workers (one thing which we need to be mindful is the distance of warehouse and work station – so in order to make water spider more efficient warehouse should be closer to the workstation for faster replenishment) Wait time – Since water spider is in charge of moving Kanban card so they know exactly how much and when material are required and hence the wait time for workers are reduce Inventory- Water spider follow the concept of JIT hence inventory pile up is out of question Who could be a water spider? At time some organization deploy junior resource as water spider but instead of that Water spider should be some with good experience and knowledgeable (SMEs) who knows in and out about the workstation they are assigned to Eg- SMEs – or checker in service industry could be termed as water spider

FMEA (Failure mode effect analysis) is a mechanism to identify step by step all possible failure modes present in a process/design/service/software so on. On the basis of three components “Severity, Occurrence and Detection” RPN (Risk Priority Number) is calculated which is product of S*O*D. Higher the RPN high is the risk involved in the step (general perception). Basis on this, there is concept of threshold RPN used by many organization. Some of the common way of identification of threshold RPN: Normally S,O and D are graded on a 10 scale metrics (where 1 is lowest and 10 is highest). So RPN can vary between 1 to 1000. Most common way of determining threshold RPN is: 1) Top 20% of the RPNs 2) Any RPN which is greater than a certain score, So on At time due to the concept of threshold RPN could be counterproductive. Eg: Process Step Severity Score Occurrence Score Detection Score RPN Step 1 6 4 4 96 Step 2 4 8 4 128 Step 3 2 8 6 96 If we consider RPN of the process steps it seems that Step 2 has higher risk and step 1 and 3 are at same risk levels, and process will put efforts in creating Mitigation and control plans for step 2 first. This is an example of disadvantages of using “threshold RPN”. But if we review closely we will find that step 1 carries highest “Severity” and miss in step 1 might leads to greater loss to organization in comparison to step 2 &3. So step 1 should proceed over step 2 &3 irrespective of lower RPN Risk Mitigation approach other than FMEA FMEA is currently not a part of ISO 14971:2019 (ISO standard for medical devices). As at time its confusing whether PFMEA/DFMEA/SFMEA is required or how each of them is integrated. Secondly, for a process there might not be only failure mode responsible for risk, (could be others especially in Medical devices) One of the better Risk Mitigation method suggested by ISO 14971 is “Risk Management System”, which is a six steps process. It starts with “Establishing of Risk Framework” (risk management process, RACI, creating a live document) 2nd step is Risk Assessment (focusing on Risk identification, analysis and evaluation- these risks could be from Business/Customer/Market so on) 3rd step is creating Risk Control document (highlighting the control measures) 4th is Risk Acceptance (there are always residual risk which the process accepts despite control plans in place, which is not a requirement in FMEA) 5th is Control Plan Review (the entire process is not reviewed always rather the control plans created are reviewed periodically) Final 6th step is Production Information (any misses, finding of any audits are the feeds for the risk management process)

Class Imbalance refers when the class (es) is/are skewed or baised (or not in proportion). For eg.- If we want to compare the performance of boys against girls in a Higher Secondary School, but if the boys: girls ratio is not 1:1 then its imbalance (eg- 5:1, or 1:5 so on) Class imbalance could be minimal or maximum (higher the class imbalance less is the accuracy of prediction) Class Imbalance impacting predictive learning using simple example Eg- Let’s take example a school want to create a predictive model to understand student performance based on their mother tongue. Assume there are 600 students mother tongue language is “A”, 200 mother tongue language is “B”, 150 speaks language “C” and 50 mother tongue language is “D” In the above scenario there are 4 classes but the frequencies are different leading to class imbalance (since the above example has more than 2 classes hence it could be referred as multi-class imbalance). When any predictive model is created one of the basic assumptions is all the classes have equal frequencies. But in the above scenario since there is class imbalance it leads to poor predictive model (especially for language D as it has the least frequency) Most Common place where we find class imbalance are: 1) Banking sector – Fraud detection (majority of the transactions are genuine and only a few are fraudulent) 2) Spams- Only a few emails are spam out of the total email transactions 3) Brand Loyalty – very less persons (in current scenario) are brand loyal on account of high options 4) Manufacturing- Only a few cars out of a lot has some performance issue

In parametric statistics, when we have to compare means of two samples there are several test where 2 samples are dependent and independent 2 Sample T Test are used to compare the means of 2 sample which are not dependent on each other and have equal variance (determined by F test) which is normally distributed. Example: If we want to compare performance of two team (sales performance) Comparing the runs scored by two different team Outcome of a drug testing on 2 independent groups Wherein Paired Sample T test is performed to compare means of 2 samples which are dependent (paired) which is normally distributed. Two means could be : Comparison pre and post-performance different times Comparison performance due to change in conditions so on Examples: Comparing the sales pre and post advertisement Comparing performance of employee per and post refresher training Person’s health before and after a treatment Benefits of using Paired T test Paired T test is a powerful tool due to the below reasons: 1) Since the sample used before and after are the same hence eliminated variation between the samples Same set of employee before and after training Same bike before and after servicing Same product before and after a marketing campaign 2) Its best tool to measure the effectiveness of some factor on a sample, industries where it could be best in use are: Service Sector – measure performance of people before and after a training Sales – Effectiveness of awareness program/advertisement (pre and post sales value and volume) Pharma- effectiveness of medicine on patient’s health Automobile – Efficiency of automobile pre and post changing fuel efficiency So, we can find its utility in all fields where ever comparative study is required on a paired Sample 3) Increase in the degree of freedom, in 2 sample T-Test df is “n1+n1-2 : as there are two different samples”, But in Paired it n-1 as the sample are same (decrease in df required high t power) 4) Time and cost is minimized as the sample size remains the same (unlike 2 sample T test wherein 2 independent samples are collated 5) The outcome of the two groups are co-related

In inferential statistics, Null and Alternate Hypothesis are defined (Ho- Null and Ha- Alternate) to conduct hypothesis testing. Null Hypothesis – states population parameters or commonly known facts Alternate Hypothesis – is the statement that one wants to prove using statistics (normally one defines alternate hypothesis first) Post defining hypothesis a sample is selected from population for the test and based on the p value either Null is accepted or Fail to accept Null Hypothesis P value < significance value (α) –Fail to accept Null hypothesis P value >= significance value (α) –Accept Null hypothesis At time due to selection of wrong samples result get impacted which is classified as Type 1 and Type 2 error So Type 1 error is normally termed as α error and Type 2 as β error In normal circumstance probabilities (significance level) are 5% and 10-20% but at time these % are altered Examples when probabilities (significance level) are very less even than 1% 1) Depends on the type of industry where one want to reduce the error –where high precision are required: a. Pharma industry b. Aviation industry c. Metro-logy products d. Ophthalmic lenses e. Hospitals etc 2) When population data is easy to collect Reducing the significance levels requires high effort (in collecting large sample) and high cost. Hence low significance level is not advisable for businesses where high precision are not required Decreasing significance level increases the chances of type 2 errors. Hence if significance levels are selected ensure that there sample size is very large to avoid errors (high sample size increase the power that is 1-β)

In Analyze phase of DMAIC, Xs’ (or causes) are identified basis different methodology of which Brainstorming is widely used. Post identification of X’s prioritization of X’s are done, where impact of X’s are studied over Y (effect). Prioritization could be performed by 2 ways “Data door or Process Door” Data door- For all the X’s where data is quantitative in nature (which can be measured). Data door method is taken. Commonly used tool for Data door approach are: Graphical Analysis: Trend Chart, Run Chats, Control Chats, Histogram etc Hypothesis Testing: Annova, Regression, Multiple Regression, Chi Sq etc For eg: if Low Accuracy is the Y for the project then potential Xs’ qualified for data door could be: “Accuracy score between employees, Volume arriving pattern vs Accuracy, Transactions type with respect to accuracy” Process door- At time there are X’s for which data is more of qualitative in nature (knowledge based upon Process experience) Process Door method is adopted. Commonly used tools are “Process Maps, Fish bone Diagram, RCA etc” For eg: if Low Accuracy is the Y for the project then potential Xs’ qualified for Process door could be: “Learning Curve of Employee, Work Environment impact on Accuracy etc”