How ML models handle large amounts of data and make decisions?

One the large millions moving already it is It shows that the zero really shows that the eight hours of the previous terms have no empire impact on the value of the cities So all values are zero that it means it is the white nice the citizens completely random So if the Syrians are completely random so we cannot do a prediction on that series that will take up mean value off ah mean value of the cities as a prediction for any bottoms is data So we come off it that this is ah real-time returns will be coming with an administration that they’re in the station it and random later.


It is Ah protection is not possible here So we take the mean value as a registered If I no So after that they’re using Mom after making that an immortal B Kumble affection for the next 10 weeks using their predict function So this as expected the Syriza predictive value off mean value that is 205 787 for all the time Although 10 predictions and the very high standard division that is 93 No I like 36,000 So prediction is like within the first gender divisions like the Blue Blazers the forecast and that you are in the range of the standard division Plus you were in minus toe so I DiLeo If so this is the same for the defender products like Samson baby Ah Whirlpool fridge We don’t know the other string all the tests that there’s a similar that he might be a getting 001 It’s only the first prediction is ah different 1725645 the next one After that it is coming Same.


You actually some for could be Are there like the Valley said are kind of time some kind of So finally via of ah ah the annual sales value for their different products like for Or you can suppose Samsung TV it is Prediction is to use it of 5.593 Vienna and the ranges from 12 to 29 legs It’s a big ah it’s a big range what any or any retail store chain owner But But what should be No wonder the safe Joan Ideally they should re-plan for this The syringe is also any and is this could have happened until that is uh what I’m saying.


You design sales revenue model customer loyalty So can you do market basket analysis on customer loyalty Type of a thing Focus only on the nb eight the transaction lived ratios one association to association items threshold values whether it is 20% you fix as the assured value off associated problem TVs bought bees also barred some rules on doing a very thorough job off-market basket analysis and customer loyalty And he grows this forecasting time Siri’s because you’re mixing too It’s not called for is what if he in this particular study does not bring this time series analysis and forecasting You remove that What a card show those retail analytics can A measure some brand preferences associations is Samson.


When they buy it Samsung something Is it Also they buy the other product of Samsung Or what is happening What variables contribute to loyalty Can I measure customer loyalty in terms off sales Bitch variables will be important Which of the items which of the associations are important to Do we have Ah association item one with one item Association of Goitom or three What are the things So if you dig deep into that you 12 and do a terrible job on a retail analytics project involving market basket analysis and association revealing and leading toe customer loyalty and revenue maximization I think that for that when I want to say that this item is associated with this of course.


I need to know what will be the sales forecast next time so that I can keep so many items of this when they bite this that will be required But what I’m trying to say from a capstone scope point of view you can award the time cities for custom It is an art that forecasting is not required to be required here for protections Next year I’m going to do then these are items required So if I have four refrigerators I might have something else But you are Study is any point in time I will say this is the probability I would say I don’t deal Stick toe that part Think about it initially for levels of generally that is a pattern Whenever we do Of course also be jumping techniques That’s a gentle problem.


I have my own mules off this missing data I have two strong bills Oneness in the manufacturing Say they draw the control chart’s upper limit Lower limit Two points are going outside They homogenize They removed two parts in there Brother control chart If there are to our place he removed all the points will be within control You’ll be giving put motion to herself You don’t deserve human about Missy because I removed them Why that has happened You should know You cannot remove similarly the missing items If there are not too many you don’t happily you have limited But if I interpolate by some method no matter water the matter I’m making a guess The missing items should be like this And if that guesses wrong all analysis is wrong around to the boat Infinity doesn’t happen in the morning Also you saw there won’t group when they when they have the not guard that they got a better result otherwise highly misleading yourselves You want to lead people toe Good decision If we’re not mislead them isn’t it So The missing items is a very sensitive area It must be an expert Really.


How do you deal with the missing value It’s a such going on Leastways Will you do limit that Ricard if you interpret late what is the pattern off the other observations in that record Can I really interpolate with a median value Will I make a mistake or is it robust The data itself shows a lot of fluctuations by anything I do I may be causing disturbance to the whole system so these are two strong points But my final advice to you is you are whole capstone you think about it Is one off market basket analysis and customer loyalty You focus only that part and Reiter report on Do little more inside you give Can I get some brand part separated here Can I measure the loyalty in terms of the people case what they were available from the card If you are able to write a wonderful capital project you can completely grow time series forecasting even though it is required for the future for the retail But you were doing market basket analysis because I always feel don’t try toe combine and go too many things in a project we thought The focus See you take one aspect Yes I’m going to go.


Similarly, the other group I told you about text analytics Then don’t bring Dexter and attribute rating Although so many ratings not required just focus on that You can do a very good job unless you go deeper to the sea you cannot pick up are you cannot extract purse You need to go deep so that you extract pulls the same thing Host toe Go deep into something and produce are probably you are demonstrating your ability to use analytics It could even be the other group I told that they have done the same thing Tabular display and all I gave a very clear message Business intelligence for this particular thing on Capstone Project where you still don’t do all this that have the bike me bike the same thing I told then don’t come optimization some linear programming There are nonlinear I tell you don’t have any of them completely complicate things Make things simple Simplify complex things do not complicate.

Leave a Comment