Web-based interactive demos

For tutorials and explanation



Sampling

Statistics: For Sampling/Inferencing

sampling

This widget allows you to change underlyig mean and variance of three three sources; namely; population, group 1, and group 2. The widget help is desgined to understands following concepts:
  • • Distribuation of means: By sampling multiple time from selected groups (+ population), it can be shown how distributation of means depends on the variability (variance of the original source)
  • • Confidence Interverval: By sampling multiple times from sources, it can be shown how confidence intervel depends on the smaple size (N)
ROC: Receiver Operating Characteristics

Machine Learning: For Understanding ROC

ml_roc_demo

In Machine Learning, ROC Curve is one of approches to evaluate the quality of a trained model by computing the AUC (Area under the curve) of ROC Curve. Although, it is straightforword to understand that higher the AUC, better it is, which is good enough to compare two or more models for the purpose of model selection. However, unlike accuracy, the interpretation of the value of AUC is not so intutive. The objective of this web-based interative widget is to help develop the missing intutive insight of AUC of ROC Curve. In addition, this widget helps to undertand the tuning of a threshold for a trained model, in order to accommondate the cost of misclassification of classes.
Sinusoidal Player

Singal Processing: Sinusoidal Player

Sinusoidal Player Sinusoidal wave with sounds

This widget is designed to understand the frequency of signal (specifically for audio). Although, the 'term' frequency is intitutive, and it is very easy to understand, how how frequency sounds differently than low frequency, however, the manifestation of frequency and it's sound is not that intitutive. For example, high frequency signal sounds like something at high level is constant, where is it is so high fluctuation, that we perceive it as if it is continuese. The hope is, that this widget, let's learners to understand, how high frequency is created and perceived.
Hypothesis Testing

Statistics: For Hypothesis Testing and T-test

sampling

In Statistics, Hypothesis testing is one of the tool used to decide if a claim has any scientific merits. The widget help is desgined from 'Sampling' widgted by adding addtional features to understands following concepts:
  • • Meaning of p-value: By sampling multiple time for 2 groups and testing, it can be seen, than even with fixed prior of distributation, sometimes, p-value suggestes to falsely reject/accept null hypothesis.
  • • Effect of sample size: By adjusting two groups close enough, it can be shown, if we have large sample size,t-test can detect the small difference. This is bases of computation of Power and minimum required sample size
  • • Effect/meaning of alpha (cut-off): By changing 0.01, 0.05, 0.001
Sampling

Statistics: For Sampling/Inferencing

sampling
ROC: Receiver Operating Characteristics

Machine Learning: For Understanding ROC

ml_roc_demo
Sinusoidal Player

Singal Processing: Sinusoidal Player

Sinusoidal Player Sinusoidal wave with sounds
Hypothesis Testing

Statistics: For Hypothesis Testing and T-test

sampling




List of demos

1. Statistics - Sampling and Inferencing.

2. ROC for PML Course (V1).

3. ROC for PML Course (V2).

4. ROC for PML Course (V3).

V3 includes more details and explanation of the widget

5. Sinusoidal Player.

6. Hypothesis Testing (with T-test).