-
Day 61
Box Jenkins Methodology A step by step process to identify and fitting ARIMA model hyperparameter; as the model itself has 3 main component: In R, the model will be wrote as ARIMA (data, order=c(p aka [AR], d aka [I], q aka [MA])). ARIMA rely on stationary data; throw back to Day 56, there are 4…
-
Day 60
Measuring Predictive Accuracy Raw data (only for illustration) X Y Forecast Value 1 0.3324 1 2 2.9232 2 3 1.4348 3 4 4.0073 4 5 3.7612 5 6 5.1456 6 7 8.1008 7 8 8.3195 8 9 8.4495 9 10 10.9755 10 11 12.1784 11 12 13.6671 12 13 14.6767 13 14 17.7715 14 15…
-
Day 59
Recap. Differencing vs smoothing = differencing is about removing trend from time series, by “station” the data back to its mean; smoothing in the other hand focus on removing irregular signal to uncover pattern (trend & seasonality). Tutorial 2 Question 1 (a) Compute a three-quarter moving average forecast for quarters 4 through 13. (b) Compute…
-
Day 58
Smoothing Techniques (Con’t) Regression on seasonality and trend Linear Regression VS Trend Component Multi-Linear Regression VS / to handle Seasonality Component Using dummy encoding to handle season. 3 dummy to handle quarterly, 11 dummy to handle monthly. Excel 1. Raw Data; Predict Year 6. Season Year 1 Year 2 Year 3 Year 4 Year 5…
-
Day 57
Smoothing Techniques Techniques Focus on Moving Average Stationary / Irregular Component Weighted Moving Average Exponential Moving Average Modified Moving Average Exponential Smoothing Holts Method Trend Component Linear / Quadratic / Exponential Trend Decomposition Model Seasonal + Trend Component Holts Winter Smoothing Dummy variable + Regression ps. Most of the time series forecast technique are under…
-
Day 56
Time Series 4 Components that made up time series 1. Irregular fluctuations – Does not follow any available pattern and not predictable; normally short period. Eg. Rise in the steel due the strike in the factory. 2. Cyclical – Large sine wave cycle about 8-10 year. Exhibit 4 phases (Peak → Recession → Depression →…
-
Day 55
Objective of time series Time Series Chart Excel 1. Raw Data Year Quarter 1 Quarter 2 Quarter 3 Quarter 4 1 667 631 675 699 2 739 695 751 779 3 823 795 835 875 4 931 855 939 967 2. Copy & paste again but holding “Ctrl” button3. Click on “Transpose” button 4. Copy…
-
Advanced Business Analytics and Visualization
Day 42 Day 43 Day 44 Donut Chart Day 45 (1) , 45 (2) , 45 (3) , 45 (4) , 45 (5) Bump ChartWaterfall ChartPareto ChartLollipop ChartChart Parameter Day 46 Tableau Storyboard Day 47 Exploratory Data Analysis (EDA) Day 48 Decision Tree Day 49 Linear RegressionImputationTransformation Day 50 Neural network Day 51 High performance…
-
Day 47
Exploratory Data Analysis (EDA) 1. Configure dataset Role accordingly 2. From Explore tab, drag StatExplore node3. Link nodes as shown4. Run StatExplore Result Tips Replace value in column using Replacement and Impute 1. From Modify tab, drag Replacement and Impute node2. Link nodes as shown 3. Configure Replacement node accordingly 4. in Replacement Editor for…
-
Day 45 (5)
Parameter 1. Click on “Age” to create Bins In Edit Bins [Age]2. Instead of fix size bin, select Create a New Parameter In Create Parameter3. Update the field accordingly 4. “Age (bin)” to Columns5. “Count” to Rows6. For parameter “Bin Size for Age” select Show Parameter7. At right side of pane, drag slider to modify…
-
Day 45 (4)
Lollipop Chart Tableau Note:– Overlay technique 1. “Segment” to Columns2. “Profit” to Rows3. “Segment” to Color 4. Do “Profit” again but this time Marks Circle 5. Combine 2 plot together using Dual Axis Spotfire Note:– In Combination Chart, no capability to set Color by attribute, only by series– no capability to create circle; hack using…
-
Day 45 (3)
Pareto Chart Tableau Note:– Overlay technique 1. “Sub-Category” to Columns2. “Sales” to Rows 3. Click on “Sub-Category” and select Sort…4. In the Sort Menu, Change to Sort by Field, Sort Order Descending and Field Name is Sales. 5. Do “Sales” again but this time Marks Line 6. Combine 2 plot together using Dual Axis Spotfire…
-
Day 45 (2)
Waterfall Chart Tableau Note:– Uses starting position of Grantt Bar as anchor, negative value of size to expand the bar 1. “Order Date” to Columns2. “Sales” to Rows 3. Bar chart cannot be use as it always start from “0” aka bottom of chart4. Grantt Bar however started at the value itself, only missing link…
-
Day 45
Bump Chart Tableau Note:– Overlay technique– Ranking and aggregation is calculated during plot initialization 1. “Order Date” to Columns2. “Sales” to Rows3. “Segment” to Color 4. Do “Sales” again but this time Marks Circle 5. Combine 2 plot together using Dual Axis Spotfire Note:– Can be done using line chart (w. Marker) or scatter plot…
Starting off as a muggle that naïve to the Math's and Data Science world.
Kim 2 ML 