| Author | Message | ||
Doosukelta Junior Artist Username: Doosukelta Post Number: 829 Registered: 04-2016 Posted From: 79.243.174.161 Rating: N/A |
oh nice to know b.sc mat aa msc kuda chsava uncle,.....present programmer or developer or teser.... idi moderator cheta cheripiveyabadindi... | ||
Andhravodu Moderator Username: Andhravodu Post Number: 11570 Registered: 07-2015 Rating: N/A |
 guess correct ye. But weather and forecasting ante baga pichi, konta understanding undi | ||
Bharateeyudu Side Hero Username: Bharateeyudu Post Number: 5815 Registered: 03-2016 Posted From: 117.195.229.32 Rating: N/A |
saaru meeru mathsaaa... mari intha active mod ante yedo keyboards ni vaththe pani anukunna | ||
Platypus Side Hero Username: Platypus Post Number: 6675 Registered: 01-2008 Posted From: 82.13.134.86 Rating: N/A |
https://cloud.google.com/ml-engine/ | ||
Platypus Side Hero Username: Platypus Post Number: 6674 Registered: 01-2008 Posted From: 82.13.134.86 Rating: N/A |
Use the power of cloud. Train the Google ML with sample data and then feed the rest of it. You can get away with a free tier for this. | ||
Andhravodu Moderator Username: Andhravodu Post Number: 11567 Registered: 07-2015 Rating: N/A |
supercomputers routinely churn hundreds of thousands or more points for weather forecasting. Ecwmf uses thousands more computers than gfs. One article on how much computing power is present and how much they spend https://www.scientificamerican.com/article/are-europeans-bet ter-than-americans-at-forecasting-storms1/ | ||
Doosukelta Junior Artist Username: Doosukelta Post Number: 828 Registered: 04-2016 Posted From: 79.243.174.161 Rating: N/A |
think not...beause...lot of variables unnai...annitni include chesi forecast cheste 95% probability tho predctions vastunnayi...like eeroju particular place lo rain padocchu ani 12-24 hours advance ga cheppocchu ....but the output is in prbabilistic scale....like based on the pesent climatic conditions in a city(temp,humidity etc etc) and satellite image of the cloud pattern, wind direction...there's 90% chance of occurrence of so and so intensity rainfall .....or 70% probability etc .... nenu alane oo model create cheyyali....but first is to drive the trend and change in trend.....idi chestee 80% work done....and prediction matram i can try and give some model which may predict long term variation...based on trend in past...done deal for 6 months... idi moderator cheta cheripiveyabadindi... | ||
Andhravodu Moderator Username: Andhravodu Post Number: 11566 Registered: 07-2015 Rating: N/A |
gor stiff like these, 175k data points won't help at all. Spread out over 30 years, you will be able to mine, but the output will be no better than a guess | ||
Doosukelta Junior Artist Username: Doosukelta Post Number: 827 Registered: 04-2016 Posted From: 79.243.174.161 Rating: N/A |
Present how rainfall pattern changed in years and how it may in future... Alane discharge in rivers past lo yela undi ippud yela undi trend chusi future prediction ki oo model... But it must keep on updating as data is flowing continuously today yesterday data include cheskunta update avutu undali... So first trend chudali next predict cheyali.... Ila oo model create cheyali.. And BTW rainfall ki dependent variables chala untai like humidity temp etc etc Naku konni months time undi... idi moderator cheta cheripiveyabadindi... | ||
Doosukelta Junior Artist Username: Doosukelta Post Number: 826 Registered: 04-2016 Posted From: 79.243.174.161 Rating: N/A |
Thanks sarang can you give me online courses or softwares that can be learned for it... Please ..TIA idi moderator cheta cheripiveyabadindi... | ||
Dreameronaroll Junior Artist Username: Dreameronaroll Post Number: 47 Registered: 08-2015 Posted From: 183.82.131.36 Rating: N/A |
Try time series analysis with variations. YOu can use Excel for this https://www.youtube.com/watch?v=gHdYEZA50KE The one and only Powerstar!!! | ||
Saarang Hero Username: Saarang Post Number: 16047 Registered: 06-2012 Posted From: 97.126.40.189 Rating: N/A |
What exactly are you trying to predict? Its tempting to do a a simple regression for any data set...it will not work in this case...as you have time series data along with cross-sectional data. Regression only works well for cross-sectional data. Most time series data has strong auto-correlation (i am sure weather data too has this), i.e. the value of variable you are trying to predict in time period t-1 will have a strong correlation with value in time period t. So, the variable is predictive of itself in time more than other variables. When you have this (time series and cross-sectional data), you have use models that come under panel data analysis. Ilaantivi chesi chaala rojulu ayyindhi...but study about fixed effects models and random effects models and try to apply one of them. GARCH models laantivi try cheyyochu but it will take a few months to learn and implement them. | ||
Anthamidhya Junior Artist Username: Anthamidhya Post Number: 160 Registered: 02-2017 Posted From: 76.122.106.214 Rating: N/A |
try doing regression on all variables first, then do another one with variables that have p < 0.05. now you can predict future y value by giving expected values for your variables in the equation | ||
Doosukelta Junior Artist Username: Doosukelta Post Number: 825 Registered: 04-2016 Posted From: 79.243.174.161 Rating: N/A |
ippudu nenu climate modelling cheyyali na deggara oka 12 climate variables wth past 30-40 years daily data undi......ante roughly 175,000 data points.... deenni model chesi trend chusi...variation note chesi.... future prediction cheyalante best method yenti....naku exel,c,c++ vacchu.... db lo sala mandi ee field lo unaru abatti can you direct me in this.... best way to do modeling and predicting trends and deriving relationships..... idi moderator cheta cheripiveyabadindi... |