├── README.md
├── _config.yml
└── index.md


/README.md:
--------------------------------------------------------------------------------
  1 | # 2019 Spring ML HW1 - Predict PM2.5 手把手教學 
  2 | 
  3 | [手把手投影片](https://docs.google.com/presentation/u/2/d/1TkPQoOPyDY9IzzuaVsYq1E26D1NTmi_QA9S9c-rw9K8/edit#slide=id.g5047f99cc6_0_0)
  4 | 
  5 | <h3 id="import">Import package</h3> 
  6 | 
  7 | ```python
  8 | import sys
  9 | import numpy as np
 10 | import pandas as pd
 11 | import csv
 12 | ```
 13 | 
 14 | <h3 id="read">Read in training set</h3> 
 15 | 
 16 | ```python
 17 | raw_data = np.genfromtxt(sys.argv[1], delimiter=',') ## train.csv
 18 | data = raw_data[1:,3:]
 19 | where_are_NaNs = np.isnan(data)
 20 | data[where_are_NaNs] = 0 
 21 | 
 22 | month_to_data = {}  ## Dictionary (key:month , value:data)                                  
 23 | 
 24 | for month in range(12):
 25 |     sample = np.empty(shape = (18 , 480))
 26 |     for day in range(20):
 27 |         for hour in range(24): 
 28 |             sample[:,day * 24 + hour] = data[18 * (month * 20 + day): 18 * (month * 20 + day + 1),hour]
 29 |     month_to_data[month] = sample  
 30 | ```
 31 | 
 32 | <h3 id="Preprocess">Preprocess</h3> 
 33 | 
 34 | ```python
 35 | x = np.empty(shape = (12 * 471 , 18 * 9),dtype = float)
 36 | y = np.empty(shape = (12 * 471 , 1),dtype = float)
 37 | 
 38 | for month in range(12): 
 39 |     for day in range(20): 
 40 |         for hour in range(24):   
 41 |             if day == 19 and hour > 14:
 42 |                 continue  
 43 |             x[month * 471 + day * 24 + hour,:] = month_to_data[month][:,day * 24 + hour : day * 24 + hour + 9].reshape(1,-1) 
 44 |             y[month * 471 + day * 24 + hour,0] = month_to_data[month][9 ,day * 24 + hour + 9]
 45 | ```
 46 | 
 47 | <h3 id="Normalization ">Normalization </h3> 
 48 | 
 49 | ```python
 50 | mean = np.mean(x, axis = 0) 
 51 | std = np.std(x, axis = 0)
 52 | for i in range(x.shape[0]):
 53 |     for j in range(x.shape[1]):
 54 |         if not std[j] == 0 :
 55 |             x[i][j] = (x[i][j]- mean[j]) / std[j]
 56 | ```
 57 | 
 58 | <h3 id="Training">Training</h3> 
 59 | 
 60 | ```python
 61 | dim = x.shape[1] + 1 
 62 | w = np.zeros(shape = (dim, 1 ))
 63 | x = np.concatenate((np.ones((x.shape[0], 1 )), x) , axis = 1).astype(float)
 64 | learning_rate = np.array([[200]] * dim)
 65 | adagrad_sum = np.zeros(shape = (dim, 1 ))
 66 |  
 67 | for T in range(10000):
 68 |     if(T % 500 == 0 ):
 69 |         print("T=",T)
 70 |         print("Loss:",np.power(np.sum(np.power(x.dot(w) - y, 2 ))/ x.shape[0],0.5))
 71 |     gradient = (-2) * np.transpose(x).dot(y-x.dot(w))
 72 |     adagrad_sum += gradient ** 2
 73 |     w = w - learning_rate * gradient / (np.sqrt(adagrad_sum) + 0.0005)
 74 | 
 75 | np.save('weight.npy',w)     ## save weight
 76 | ```
 77 | 
 78 | <h3 id="Write file">Read in testing set</h3> 
 79 | 
 80 | ```python
 81 | w = np.load('weight.npy')                                   ## load weight
 82 | test_raw_data = np.genfromtxt(sys.argv[2], delimiter=',')   ## test.csv
 83 | test_data = test_raw_data[:, 2: ]
 84 | where_are_NaNs = np.isnan(test_data)
 85 | test_data[where_are_NaNs] = 0 
 86 | ``` 
 87 | 
 88 | <h3 id="Predict">Predict</h3> 
 89 | 
 90 | ```python
 91 | test_x = np.empty(shape = (240, 18 * 9),dtype = float)
 92 | 
 93 | for i in range(240):
 94 |     test_x[i,:] = test_data[18 * i : 18 * (i+1),:].reshape(1,-1) 
 95 | 
 96 | for i in range(test_x.shape[0]):        ##Normalization
 97 |     for j in range(test_x.shape[1]):
 98 |         if not std[j] == 0 :
 99 |             test_x[i][j] = (test_x[i][j]- mean[j]) / std[j]
100 | 
101 | test_x = np.concatenate((np.ones(shape = (test_x.shape[0],1)),test_x),axis = 1).astype(float)
102 | answer = test_x.dot(w)
103 | ```
104 | 
105 | <h3 id="Write file">Write file</h3> 
106 | 
107 | ```python
108 | f = open(sys.argv[3],"w")
109 | w = csv.writer(f)
110 | title = ['id','value']
111 | w.writerow(title) 
112 | for i in range(240):
113 |     content = ['id_'+str(i),answer[i][0]]
114 |     w.writerow(content) 
115 | ``` 


--------------------------------------------------------------------------------
/_config.yml:
--------------------------------------------------------------------------------
1 | theme: jekyll-theme-cayman


--------------------------------------------------------------------------------
/index.md:
--------------------------------------------------------------------------------
 1 | ## Assignment 1 - Predicting PM 2.5
 2 | 
 3 | You can use the [editor on GitHub](https://github.com/ntumlta2019/hw1/edit/master/index.md) to maintain and preview the content for your website in Markdown files.
 4 | 
 5 | Whenever you commit to this repository, GitHub Pages will run [Jekyll](https://jekyllrb.com/) to rebuild the pages in your site, from the content in your Markdown files.
 6 | 
 7 | ### Markdown
 8 | 
 9 | Markdown is a lightweight and easy-to-use syntax for styling your writing. It includes conventions for
10 | 
11 | ```markdown
12 | Syntax highlighted code block
13 | 
14 | # Header 1
15 | ## Header 2
16 | ### Header 3
17 | 
18 | - Bulleted
19 | - List
20 | 
21 | 1. Numbered
22 | 2. List
23 | 
24 | **Bold** and _Italic_ and `Code` text
25 | 
26 | [Link](url) and ![Image](src)
27 | ```
28 | 
29 | For more details see [GitHub Flavored Markdown](https://guides.github.com/features/mastering-markdown/).
30 | 
31 | ### Jekyll Themes
32 | 
33 | Your Pages site will use the layout and styles from the Jekyll theme you have selected in your [repository settings](https://github.com/ntumlta2019/hw1/settings). The name of this theme is saved in the Jekyll `_config.yml` configuration file.
34 | 
35 | ### Support or Contact
36 | 
37 | Having trouble with Pages? Check out our [documentation](https://help.github.com/categories/github-pages-basics/) or [contact support](https://github.com/contact) and we’ll help you sort it out.
38 | 


--------------------------------------------------------------------------------