+ All Categories
Home > Software > Gold Mining in Data Mining

Gold Mining in Data Mining

Date post: 15-Jul-2015
Category:
Upload: ahmad-kaifi
View: 267 times
Download: 2 times
Share this document with a friend
Popular Tags:
18
Gold Mining in Data Mining By Ahmad Kaifi Hassan Althobaiti
Transcript
Page 1: Gold Mining in Data Mining

Gold Mining

in

Data Mining

By

Ahmad Kaifi

Hassan Althobaiti

Page 2: Gold Mining in Data Mining

Outline

Introduction

Motivation

Approach

Result

Demo (using Shinny)

Limitations

Page 3: Gold Mining in Data Mining

Introduction

Suppose you have a dataset that contains

gold prices from 1980 until now.

What kind of technique are you going to use?

Page 4: Gold Mining in Data Mining

Forecasting

Is the idea of using old information in order to

predict future information.

“Prediction is very difficult,

especially if it's about the future.”

Niels Bohr

Page 5: Gold Mining in Data Mining

Principles of Forecasting

Forecasts are rarely perfect

Forecasts are more accurate for grouped data than for individual items

Forecast are more accurate for shorter than longer time periods

Page 6: Gold Mining in Data Mining

Motivation

Since the forecasting is always wrong, our

motivation is to test all the different algorithms

available in “Forecast” package in three

different time periods based on the duration

and the trend from 1980 until now in order to

determine some algorithms that can mimic the

actual data and give more accurate results.

Page 7: Gold Mining in Data Mining

Approach

MASE

Page 8: Gold Mining in Data Mining

Results Of Duration Test

FF-1980 MASE

stlf 1.19

ses 1.26

arfima 0.78

auto.arima 1.20

FF-2007 MASE

stlf 0.30

ses 0.46

arfima 0.43

auto.arima 0.70

FF-2011 MASE

stlf 0.41

ses 0.51

arfima 1.49

auto.arima 0.35

Page 9: Gold Mining in Data Mining

Results Of Duration Test

FF-1980 MASE

meanf 9.67

stlf 1.19

arfima 0.78

HoltWinters 1.49

FF-2007 MASE

meanf 0.22

stlf 0.30

arfima 0.43

HoltWinters 0.15

FF-2011 MASE

meanf 1.99

stlf 0.44

arfima 1.49

HoltWinters 0.60

Page 10: Gold Mining in Data Mining

Results Of Duration Test

FF-1980 MASE

thetaf 1.31

holt 1.26

hw 0.78

auto.arima 1.20

FF-2007 MASE

thetaf 0.61

holt 0.65

hw 0.65

auto.arima 0.70

FF-2011 MASE

theta 0.33

holt 0.25

hw 0.32

auto.arima 0.35

Page 11: Gold Mining in Data Mining

Observation

For long-term prediction, the ARFIMA model is

the best.

For short-term prediction, the HoltWinters

model is more accurate.

Page 12: Gold Mining in Data Mining
Page 13: Gold Mining in Data Mining

Result Where GP Trend

Remains Constant

1980-1995 MASE

ses 0.36

nnetar 1.59

tbats 1.58

auto.arima 1.13

Page 14: Gold Mining in Data Mining
Page 15: Gold Mining in Data Mining

Result Where GP is Decreasing

1980-2008 MASE

snaive 0.36

croston 1.59

stlf 1.58

nnetar 1.13

Page 16: Gold Mining in Data Mining
Page 17: Gold Mining in Data Mining

Result Where GP in Peak

1980-2011 MASE

thetaf 1.83

naive 1.81

ses 1.80

arfima 1.89

Most of the algorithms are unpredictable.

Page 18: Gold Mining in Data Mining

Demo


Recommended