Anyone know how to do weather prediction with machine learning with website ?

[Chaitanya kc]

I am searching everywhere for the help I can't find any most of stuff I found is related to only prediction part only I need to to deploy in website. Please share any links regarding this project . I am student doing my final year project I can't figure this !

[jj9987]

You can't do machine learning in a website per se. Why do you need to deploy it in website?

 

If you want to visualize, set up a backend, which does the machine learning, and frontend, which handles the visualization.

[Franck]

22 minutes ago, Chaitanya kc said:

I am student doing my final year project I can't figure this !

Well meteorologist can't figure it out either honestly. If you manage to figure it out you will make ALOT of money. Good luck

[straight_stewie]

2 hours ago, Chaitanya kc said:

I need to to deploy in website

What is specifically meant by this?

Do you mean "I need to be able to share the results of the model"?

Do you mean "I need to be able to share an executable version of the model"?

Do you mean "I need the model to execute in browser"?

Do you mean "I need to get data from a website to build the model"?

[reniat]

Machine learning the weather to any degree of accuracy is actually an unsolved problem afaik. Weather is affected by the surrounding weather, so making predictions kinda ends up looking like chaos theory. There are so many variables that we don't have the ability to just put the weather data into a traditional model.

 

Maybe somehow the data can be simplified, and then learned on, resulting in a still pretty inaccurate but better than human prediction weather forecast?

[Chaitanya kc]

9 hours ago, straight_stewie said:

What is specifically meant by this?

Do you mean "I need to be able to share the results of the model"?

Do you mean "I need to be able to share an executable version of the model"?

Do you mean "I need the model to execute in browser"?

Do you mean "I need to get data from a website to build the model"?

I am working on able share the results from the model on a website 

[Chaitanya kc]

7 hours ago, reniat said:

Machine learning the weather to any degree of accuracy is actually an unsolved problem afaik. Weather is affected by the surrounding weather, so making predictions kinda ends up looking like chaos theory. There are so many variables that we don't have the ability to just put the weather data into a traditional model.

 

Maybe somehow the data can be simplified, and then learned on, resulting in a still pretty inaccurate but better than human prediction weather forecast?

My project is focused on deploying machine learning model to predict weather on web server 

[Chaitanya kc]

9 hours ago, jj9987 said:

You can't do machine learning in a website per se. Why do you need to deploy it in website?

 

If you want to visualize, set up a backend, which does the machine learning, and frontend, which handles the visualization.

My goal is to run python scripts on web server with previously developed machine learning model like flask framework 

[Chaitanya kc]

5 minutes ago, Chaitanya kc said:

My project is focused on deploying machine learning model to predict weather on web server 

Like gis website 

Example : https://mangomap.com

[Chaitanya kc]

9 hours ago, Franck said:

Well meteorologist can't figure it out either honestly. If you manage to figure it out you will make ALOT of money. Good luck

Thanks 

[vorticalbox]

15 hours ago, reniat said:

Machine learning the weather to any degree of accuracy is actually an unsolved problem afaik. Weather is affected by the surrounding weather, so making predictions kinda ends up looking like chaos theory. There are so many variables that we don't have the ability to just put the weather data into a traditional model.

 

Maybe somehow the data can be simplified, and then learned on, resulting in a still pretty inaccurate but better than human prediction weather forecast?

could always go on historical data right? in april in tends to rain a lot, looking back across n number of years to seeing if it rained 

[FlappyBoobs]

A supercomputer that can provide 16 day "accuracy" (but really it is only about 5 days accurate at best) in predictions has 8.4 Petaflops of power.

 

By comparison the i9-7980XE has a single teraflop of power. Put simply, you cannot do what you want to do to with the computing power available to you on a webserver.

[reniat]

3 hours ago, vorticalbox said:

could always go on historical data right? in april in tends to rain a lot, looking back across n number of years to seeing if it rained 

yeah but that's not likely gonna be more accurate than current techniques...

[Franck]

12 minutes ago, reniat said:

yeah but that's not likely gonna be more accurate than current techniques...

Well i used to deal with all US/CAN weather stations data to actually try to get a trend so we can estimate the average dry bulb/ wet bulb/ grains / HR which are quite critical in thermodynamics performance calculation for outdoor equipment such as coils / industrial AC. Old data trends are still very very variable in more than majority of stations datas.

[reniat]

17 minutes ago, Franck said:

Well i used to deal with all US/CAN weather stations data to actually try to get a trend so we can estimate the average dry bulb/ wet bulb/ grains / HR which are quite critical in thermodynamics performance calculation for outdoor equipment such as coils / industrial AC. Old data trends are still very very variable in more than majority of stations datas.

That's pretty cool. what was the period for those averages? like how precise could you get your estimates? by day, week, hour?

[Franck]

55 minutes ago, reniat said:

That's pretty cool. what was the period for those averages? like how precise could you get your estimates? by day, week, hour?

Weather station depending on the station has data per hour or per day. You also have pre calculated monthly and yearly average and other data such as days it rained, duration, quantity. We were mostly interested by extremities average of the location. lowest and highest average temperature, and humidity so we could estimate the most cooling or heating we can achieve at these worst condition to then balance with the building management system presets and figure out all building internal data that affect those to pick the best equipment you can so we can built it to size perfectly for each clients.

 

Typically we were looking the 4 seasons individually for each locations. Most hot dry, most hot wet, most cold dry, most cold wet etc..

 

We had data down to the 1960 but there were very few data in that decade (mostly Dry and wet bulb, bar and wind direction/speed). Nowaday you have over 50 variables per hour on busy stations such as those in airports (most main ones are located in airports btw). We ended up using mostly last 10 years as a base and users could override to choose only last year straight instead of estimate / probability.

[Sauron]

On 2/27/2019 at 7:03 PM, jj9987 said:

You can't do machine learning in a website per se.

Well, technically you could - you just need to implement the learning part in the backend and then upload a training set. I wouldn't necessarily recommend it, but it's possible.

On 2/28/2019 at 4:28 AM, Chaitanya kc said:

My goal is to run python scripts on web server with previously developed machine learning model like flask framework 

What's the issue? Are you having trouble with python, with the web server or something else?