Dhruva

According to me a prototype is a design that one can think of that how an end product should look like. whereas provotype looks an enhancement procedure from end user perspective and considers dmaic/dmadv techniques to continuously build a refined prototype. i think Apple Inc involves lot of time in provotypes and make user friendly hardware and software, continuously fix bugs in ios releases. i think PESTLE plays a bigger role in conducting a provotype as the external forces can influence a prototype

Hammock is a swing tied at two opposite corners . Therefore by definition it hangs between two corners. In a project management instance, it would mean a task that that hangs between two dates. In a six sigma perspective it would mean to carry ‘x’ for a period of time before applying f(x) so that it becomes a y. For example in Theory of constraints we studied to increase the capacity of existing resources or deploy more capacity at non constraints. So if a resource who is a non constraint needs to work extra days to build in capacity, needs to be pid wages and acts as a hammock or also a supporting link between two dates.

Project Premortem is the term that refers to detect or control the occurrences of an unexpected failure that may arrive in the future and may lead to death of a project. it is more of an envisaging technique to forecast future milestones in a project and act earlier than being late. it can be explained with an example of a company who hires a vendor to build a software code as this company has provided a solution for the company and is a trademark intelligent software. if the supplier does not understand the requirements correctly and do not ask questions, there could be a case where timelines can be extended for the project, scope creep can appear, and also at times, the functionality can fail due to regulatory compliance. The limitations of this can be observed in bigger projects which have too many occurrences of milestone plan; eg Delhi Metro Project. this would mean forecasting and looking ahead far to the delivery angle and bring all risks and issues that can happen and would involve a lot of time, intelligent resources and money in planning.

AEIOU are the vowels in alphabet but can be an end to end story in design thinking. Similar to vowels forming echolia in language , this framework forms the base of design thinking. A- Activities E- Environments I - Interactions O- Objects U - Users In a software development world, if one has to develop one feature, can be explained by design thinking AEIOU framework For example the feature that needs to be built is a bot for case management tool. Activities involved will be built the bot and test it Environment will mean whether to build in a live environment or a test environment Interactions would involve Developer to Bot, Tester to Bot and User to Bot or Bot to User Objects would include BOT creation in Ui path or Blue Prism Users would be the parties involved- Business Analysts, Developers, Testers and End Users Now what did this whole framework tell us? It definitely helps develop an operating model framework in our minds and gives a basic idea on how will an wnd product work.

While a project charter is traditinally accepted form of accepting a project with stakeholder roles and the problem/goal statement, BOSCARD is more practical way of looking at things from agility angle. In a similar way to Project Charter, BOSCARD also defines background and objectives including the scope. However , BOSCARD also evaluates the constraints, Risks, Assumptions and Deliverables which is more like a RAID log in agile world and more realistic and practical. The biggest problem with project charter is the practical application becomes difficult because of not covering the assumptions, risk and Issues in a specific manner but is embedded in various headers. In my opinion BOSCARD covers end to end details to a project with all the parties involved and spends more time in attention to details.this is a next gen way of working style.

Float is the difference of Latest start time/date and Earliest start time/date in a critical path method. In other words , one first needs to determine earliest start time . Latest start time is achieved by subtracting duration from latest finish time (right to left) Critical path method is where earliest start time and latest finish time are the same. Now the float can help determine whether it is possible to defer a project by few days or not. If activity 1 has an earliest start time of 0 days and latest start time of 2 days, then essentially it tells one that the difference or float of 2 days can be given to the project.

According to the Pareto analysis, 80 percent of the problems are because of top 20 percent causes. Or in other words, 20 percent of the reasons cause 80 percent of the consequences. One must not be adamant to be saying that 80:20 principles means actually 80:20, that is not correct. Realistically 70:30 also gives or portrays a better picture if one needs to eliminate maximum causes/reasons. Whether 80:20 or 70:30 or 60:40, the onus lies in fixing what has been the source of problems and must be eliminated as quickly as possible. Now, 80:20 is just a decebt indication to tell one that major problems are caused by a few reasons , however it depends on business appetite whether they need tight specs or broad specs using 90:10, 80:20, 70:30 or any other combination.

R squared determines goodness of fit or in other words is a coefficient of determination. Here we need to understand if R2 is 100% meaning it holds a value of 1, then what is inferred. This can be explained with the help of below diagram Chart 1 would show us that how closely the values in orange variable Y are dependent on blue variable X meaning a high score of R2 On the other hand, Chart 2 would show us that how near or far the values in orange variable Y are dependent on blue variable X meaning a slightly lower score of R2 Let us observe in Chart 2 where X is 6 . That is the difference between X and Y meaning that the observed values are deviated from the regression line/data set. If it was same as a value of 6 for both and all other data points holding same value corresponding to X in Y, then it means there is no variation between X and Y and these are closely tied. In reality, it would be highly impossible that the two correspond to each other and equate an R2 value of 1. Examples of such an industry would obviously not be from pathological labs but can be where data sets are whole numbers. For example X hit Z because Y asked to do it and so on. Typically high R2 would not denote a very good sign for anything as more is relied upon value of p.

It is believed that process improvements do not always contribute to change in specification’s limits. We brought in a machine learning data model to one of our projects and defined recall rates i.e. the capability of the model to identify the true sanction organisations as an interval between 90% and 98%. However, the control limits showed that within a volume of 1 million, we were with 92%and 94% percent control. Now the question is the upper specification varies slightly more than upper control limit. So there is always a scope to bring specification from 98% to 95% as one would think as a rational thinker. However, if there are reasons to keep the gap broad between upper control and upper specification limit due to regulatory requirements such as aircraft wing span in a runway, then specs must not be reduced or changed is my opinion.

It would mean severity of the process is 10(most risky process), occurrence is 10 ( always fails) and detection is 10(extremely late detection). We know from our experience that we need to focus on controlling the occurrence than detection. But here in this case if it is most risky process and occurrences are the highest and we get to know at the end, this would be a complete failure scenario and sounds unrealistic i.e. completely raw processes. We cannot expect an army person operating a drone at the border of two enemy countries to cross the border everytime he operates it. Further, let us consider an example of a pizza makeline. Once the chef puts a pizza in an oven, the desired output is a hot pizza which must meet the food and hygiene standards and customer satisfaction. Assume that everytime pizza comes out of the oven, is damaged. So from here we know occurrence is high i.e. everytime it fails. It has more severity as every damaged pizza has a mould in it and detection is also quite late. But is it 10? From a process step of baking, it is ten and yes it is a complete failure. But could it be detected earlier? Yes it could be detected earlier by naked eye or a simple examination of the oven. But it seems the processes like brick making where visibility could be low or the least or projects where one sees the final product only at the end, the detection seems to be higher or highest. However, in most cases it sounds quite unrealistic for RPN of a step to be 1000 as most processes and services have a troubleshooting mechanism in place.

There could be potentially zero correlation between known cause and effects, meaning when one variable is set to some value, the other variable may still fall or rise. We can explain it with an example. There is a male boy of 6 years with autism. As the age of the boy increases to 7, the level of autism or ASD CARS(childhood autism rating scale) may rise or fall even though there is a cause and effect assumption that autism rate will diminish as the child grows because the neurons or hyperactivity tend to settle down/burnt out with age. Further there is global data available that shows boys have more autism than girls. But still in principle or reality could be opposite due to uncollected or not analysed data. This in no way would means that if boys keep increasingly diagnosed with autis , the rate at which girls are detected would come down. That would be a wrong assumption to make. i would like to explain with another example too using the same autism spectrum disorder (ASD). In order to control the hyperactivity, there is a medicine used called Risperidone. Now if we were to give 1ml or 20 drops to a child aged 5, would mean he must be more settled than his peers with same level of ASD. If we increase the dosage to 1.5ml or to 30 drops in a day, it can be both adverse or favourable. So there is no correlation between childhood age autism and dosage of medicine.

Control limit essentially determines whether our processes are in control or out of control. To a large extent the control limits are indluenced because these are data driven. I did a process on Machine Learning where the control was 5% was acceptable. So basically 95% was agreed confidence of identifying 95 high risk people out of 100. Today in a volume of 200 , we say 5 % we can miss, which is a tolerance or control, maining 10 high risk people ot type B can be missed. Suddenly the volume dipped to 50 customers the next working day and we still missed 6 high risk people meaning 12 % loss instead of 5 %. The places and situations where whole number needs to be applied are greatly influenced. This meant that although our control limit was 90 % lower and 95 % higher, we still were 88 percent which influenced the situation. And the influencer was none other than the data.

The coefficient of correlation is hardly perceived as a perfect +1 or -1. The obvious reason is the complexity of the formula used might not bring exact +1 or -1. The other reason might be that i can think one of my ownself. I take clonezapam(anti anxiety medicine or alpazolam) of potencity 0.25mg every night. I get a sleep of 8 hrs. So one tablet and one night sleep of 8 hrs may look like a perfect correlation. However, if i increase the potencity of 1 tab to 0.5mg, i will still sleep 8 hrs, but i think the sleep will be more sound and will make me feel more fresh the next morning. Now although it is still one tablet and one night sleep, but i cant exactly measure how much out of 100pc, was 0.25mg effective or was 0.5mg effective. And vice versa for negative correlation. Perfect example could be how much anxiety goes down by taking 0.5mg instead of 0.25mg. On the flip side, i also think that artificially we can achieve perfect correlation as used by the infertility clinics. However, if we need to analyse the depth, we will have to use several procedures to study the composition of an embryo to have it achieved perfect correlation. So in my own thoughts, perfect correlation could be visible only in picture, but the reality behind is far complex which might be near to perfect correlation but not absolute correlation.