Amazing AI Use Cases-Google Cloud ML Increases On Time Flight Using Wind Forecasting

Key Takeaways

hello all my name is krishnaik and welcome to my YouTube channel so guys as an AI practitioner where I've specifically spent more than 10 plus years of experience in the analytics industry I believe with the help of artificial intelligence we can definitely solve a lot of problem statement and the most important thing is that you really need to have a good domain knowledge today in this specific video I really want to talk about one amazing use cases regarding how Google is able to implement something related to wind forecasting and how it is basically used in an airport industry right this is quite amazing I still believe that in every every domain every technology in every industry AI is going to be getting used so let me go ahead and let me just show you this about this amazing blog that which recently Google cloud has basically published and you will be able to understand that what all things they have actually focused on what all data set they have actually focused on how did they collect the data set what was the performance metrics that they had actually seen now this is super important for any people who are working in different Technologies domain you should also start in the similar way yes irrespective of domain knowledge the problem statement May differ the problem solving uh technique may basically differ so let me go ahead and let me just share my screen quickly so this was an amazing news from Lufthansa Lufthansa increases on flight time delays by wind forecasting with the help of Google Cloud now I'll give you a two minutes time just pause the video right and just see the specific thing and just see that with the help of this wind direction why do you think in the left hand side there is a delay and on the right hand side because of the future forecasting of this when forecasting you will be able to see that the flights are able to ah you know depart at the scheduled time right so just pause the video pause the specific video and just think about it before that I go ahead I really want to quickly announce if you really want to know that how we can basically solve this kind of use cases definitely we are coming up with this amazing data science Industry ready projects which is basically which has basically started from September 24th but yes we have just completed the introduction in class I would suggest come over here try to understand how do we solve our end-to-end data science projects by using various clouds like AWS Azure and gcp along with this we are also coming up with this full stack data science bootcamp with the job guaranteed program that is bootcamp 2.0 and again here the focus will be that we'll try to learn data science from the scratch now let me go ahead and let me talk about this specific use case now here you can see that the forecasting that is the wind forecasting that is done by the Google Cloud ml right what has happened is that it has basically helped the flight to depart at the scheduled time so before this now you here you can basically see that without the forecasting now there is a concept of Against the Wind and for the wind you know with respect to any airport industry now because of this on the left hand side you will be able to see that since they were not able to predict it correctly here you can see that the flight is getting delayed and they are passing through this and the wind is basically coming from this direction to this direction but when you fly it right you basically go against the wind right so this is a kind of a small domain knowledge that you specific need to refer it and over here all the information regarding this that how it has basically solved how the collecting and the preparing of the data set has basically happened everything is there I just want to show you that what all things has basically been used the entire wind forecasting here you can see that the goal of Lufthansa and Google Cloud project was to project the bisc win for jury clothing airport using deep learning based model approaches then to see if the prediction surprises the internal heuristic driven solution and cause the ease of use and practicality of the deep learning approach right so here you can basically see that what all things they have actually focused on they actually try to collect a data from past five years the collected data was next subjected to an extensive cleaning and feature Engineering Process using vertex AI workbench so this is a workbench that is available in Google but let's say that you are also starting right some or the other way you're also going to do the same process manually right in order to prepare the final data set for training the cleaning phase included steps to drop the features or rows that contains too many missing values a failed statistical test for entropy Etc since the direction of the wind is circular between 0 and 360 degree this column of feature was replaced with two features so here you can see with respect to cosine and sine embedding is also being used now how did they came up with all these particular techniques it is because of the specific domain knowledge now here you can see the data set was then flattened such that the columns contain all the relevant features and send some measurements for all the weather station and all so here also you'll be able to see the what all features they have actually used they try to derive features like wind direction speed pressure temperature humidity more at a 10 minute resolution so 10 minutes so the time interval based on that they were able to do the forecasting of this win now see the model will be basically saying now if I probably start over here the model will be saying that okay the wind is basically going to come in this way so before that all the flights have basically departed then when the wind started coming opposite to the wind direction because it will help uh you know the flay uh the flight to basically depart or flight to basically fly it is basically going in the next Direction but because of this here on the left hand side you can see because of that the flight Direction the flight departure Direction has not been changed uh so because of that there is a huge delay right so this is an amazing use case and here you can basically see what kind of architecture training pipeline they have basically used manually also if you think right almost we do the same thing right so here it is and they have also used some of the parametrics over here like what are parametrics they should definitely use that is shown what is the results and next step every graph has basically same so I would suggest definitely go through this but the most and the foremost important thing is that how amazing use case is being solved at the end of the day if you have some good amount of domain knowledge any use cases in any technology in any domain that you are specifically working you don't have to worry about it you will be able to solve with the help of AI and this is what you should also start looking for whether you are working in any companies you tell your manager I found out this specific use case I can solve this specific use case at the end of the day just by creating this you know what will happen if the flight delays are not happening millions and millions of dollars will be saved once Lufthansa basically executes it they have also started executing it and just amazing results have basically happening in this article they have also said that how much percentage of money they are able to save right so this is a super important thing right so please do let me know that if you want similar kind of videos where I specifically talk about amazing use cases and all I'll be happy to talk about it and yes at the end of the day you really need to learn a different things over here because I cannot just keep on explaining each and everything because it may also be something very new to me with respect to the domain knowledge but yes this is really giving a good amount of idea about it so yes this was it for my side I'll see you in the next video have a great day bye