|
| 1 | +" |
| 2 | +[[[ |
| 3 | +| graph | |
| 4 | +graph := #(5 6 5 6 4 2 10 10 10 10 12 12 14 8 5 6 4 2 3 4 5 6) histogram. |
| 5 | +graph view canvas open |
| 6 | +]]] |
| 7 | +" |
| 8 | +Class { |
| 9 | + #name : #FluDataset, |
| 10 | + #superclass : #Object, |
| 11 | + #instVars : [ |
| 12 | + 'rawData', |
| 13 | + 'retrievalTime' |
| 14 | + ], |
| 15 | + #classInstVars : [ |
| 16 | + 'uniqueInstance' |
| 17 | + ], |
| 18 | + #category : #ComputationalDocuments |
| 19 | +} |
| 20 | + |
| 21 | +{ #category : #'instance creation' } |
| 22 | +FluDataset class >> new [ |
| 23 | + self error: 'Author is a singleton -- send uniqueInstance instead' |
| 24 | +] |
| 25 | + |
| 26 | +{ #category : #'instance creation' } |
| 27 | +FluDataset class >> reset [ |
| 28 | + <script> |
| 29 | + ^ self uniqueInstance reset. |
| 30 | +] |
| 31 | + |
| 32 | +{ #category : #'instance creation' } |
| 33 | +FluDataset class >> uniqueInstance [ |
| 34 | + ^ uniqueInstance ifNil: [ uniqueInstance := super new ] |
| 35 | +] |
| 36 | + |
| 37 | +{ #category : #accessing } |
| 38 | +FluDataset >> asInterpretedDataFrame [ |
| 39 | + | data | |
| 40 | + data := self asRawDataFrame. |
| 41 | + data column: 'week' |
| 42 | + put: ((data column: 'week') collect: |
| 43 | + [ :each | Week year: (each first: 4) asNumber week: (each last: 2) asNumber ]). |
| 44 | + #('inc' 'inc_low' 'inc_up' 'inc100' 'inc100_low' 'inc100_up') do: |
| 45 | + [ :c | data column: c put: ((data column: c) collect: #asNumber) ]. |
| 46 | + ^ data |
| 47 | + |
| 48 | +] |
| 49 | + |
| 50 | +{ #category : #accessing } |
| 51 | +FluDataset >> asRawDataFrame [ |
| 52 | + | lines data | |
| 53 | + lines := rawData lines. |
| 54 | + data := DataFrame fromRows: ((lines allButFirst: 2) collect: [ :each | each splitOn: ',' ]). |
| 55 | + data columnNames: (lines second splitOn: ','). |
| 56 | + ^ data |
| 57 | + |
| 58 | +] |
| 59 | + |
| 60 | +{ #category : #accessing } |
| 61 | +FluDataset >> extractColumns: names [ |
| 62 | + | lines columnNames indices data | |
| 63 | + lines := rawData lines. |
| 64 | + columnNames := lines second splitOn: ','. |
| 65 | + indices := names collect: [ :each | columnNames indexOf: each ]. |
| 66 | + data := Array2D rows: lines size - 2 columns: names size. |
| 67 | + (lines allButFirst: 2) withIndexDo: |
| 68 | + [ :line :i | |
| 69 | + | values | |
| 70 | + values := line splitOn: ','. |
| 71 | + indices withIndexDo: [ :j :jj | data at: i at: jj put: (values at: j) ] ]. |
| 72 | + ^ data |
| 73 | +] |
| 74 | + |
| 75 | +{ #category : #initialization } |
| 76 | +FluDataset >> initialize [ |
| 77 | + rawData := 'http://www.sentiweb.fr/datasets/incidence-PAY-3.csv' asZnUrl retrieveContents. |
| 78 | + retrievalTime := DateAndTime now. |
| 79 | +] |
| 80 | + |
| 81 | +{ #category : #accessing } |
| 82 | +FluDataset >> rawData [ |
| 83 | + ^rawData |
| 84 | + |
| 85 | +] |
0 commit comments