Apply Spark Platform to Predict the Model of Temperature in Huafan University
| Autor: | LING, WEI-FENG, 凌偉峰 |
|---|---|
| Rok vydání: | 2017 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 105 There has been increasing attention paid to the issue of global warming. According to the National Oceanic and Atmospheric Administration report on global warming in 2015, global average temperatures continue to rise and break a record set in 2014. It was the warmest year since 1880. Taiwan was affected by a Cold current in January 2016, and even just 500 meters above sea level that temperatures fell below 0 degrees Celsius. The Taipei meteorological station repeatedly detected in high temperature more than 38 degrees Celsius from June to July in the same year, so it impacts heavily on Taiwan's ecosystem. The Huafan University Sustainable Development Research Center of Slope Land has built weather station and automatic Inclinometer stations. This thesis uses the information of weather station and automatic Inclinometer for analysis after processed. The aim of this study is to find the temperature prediction model that using R language for Linear Regression, Neural Network, and Generalized Linear Models of Normal Distribution on Apache Spark platform. First, datasets are normalized. Then, the Linear Regression and Normal Distribution in Generalized Linear Models are applied to obtain the root mean square error. These results are 2.02561 and 2.166586, respectively, The Skip parameters of the Neural Network (whether Neural Network with hidden layer or not) is set as True and False, and the results of root mean square error are 2.06964 and 1.298371, respectively. From the simulation results, the obtained root mean square error of the Neural Network with hidden layer has the best results. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|