Many researchers are now convinced that a good proportion of the benefit derived from real medication is received from the placebo or “halo” effect. Since everyone, including the doctor, knows that extensive testing goes into new drugs, when one is released for use, the doctor expects it to work, the patient expects it to work [...]
Mar
31
MODEL EVALUATION TECHNIQUES FOR THE CLASSIFICATION TASK
Perhaps the most widespread usage of supervised data mining involves the classification task. Recall that in classification, there is a target categorical variable. The data mining model examines a large set of records, each record containing information on the target variable as well as a set of input or predictor variables. The analyst would like [...]
It is well known that physicians regularly use placebos – sugar pills or pills with absolutely no real medical power. The patients, however, are told that the pills are powerful medicaments. Countless studies have proved the high effectiveness of these “mind only” medications.
In a 1979 study, patients with severely bleeding ulcers were split into two [...]
Mar
30
MODEL EVALUATION TECHNIQUES FOR THE ESTIMATION AND PREDICTION TASKS
For estimation and prediction models, which employ supervised methods, we are provided with both the estimated (or predicted) value y of the numeric target variable and the actual value y. Therefore, a natural measure to assess model adequacy is to examine the estimation error, or residual, |y ? y|. Since the average residual is always [...]
Gallwey, therefore, taught his players to engage, or distract, the verbal Self 1 during play, by describing external events. They would say “bounce” when the ball bounced, or “hit” when it struck the racket. They alternatively would be told to say the words of a song. These distractions, left brain activities, allowed the right brain [...]
Mar
29
MODEL EVALUATION TECHNIQUES FOR THE DESCRIPTION TASK
In Chapter 3 we learned how to apply exploratory data analysis (EDA) to learn about the salient characteristics of a data set. EDA represents a popular and powerful technique for applying the descriptive task of data mining. On the other hand, because descriptive techniques make no classifications,
Tim Gallwey in his best selling book “The Inner Game of Tennis”, showed how visualisation can be much more effective than verbal instruction. As a tennis Pro, he became aware that each pupil’s mind seemed to contain two entities. A Self 1 who observed and commented on the play, and a Self 2 who actually [...]
Mar
28
MODEL EVALUATION TECHNIQUES
MODEL EVALUATION TECHNIQUES FOR THE DESCRIPTION TASK
MODEL EVALUATION TECHNIQUES FOR THE ESTIMATION AND
PREDICTION TASKS
MODEL EVALUATION TECHNIQUES FOR THE CLASSIFICATION TASK
ERROR RATE, FALSE POSITIVES, AND FALSE NEGATIVES
MISCLASSIFICATION COST ADJUSTMENT TO REFLECT REAL-WORLD CONCERNS
We act not according to what things really are – but according to what we expect them to be: believe them to be: imagine them to be.
“Imagination,” said Napoleon, “rules the world”.
He should have known, for he actually rehearsed every battle he ever fought weeks before the event in his mind. Going over his own [...]
Mar
27
LOCAL PATTERNS VERSUS GLOBAL MODELS
Finally, data analysts need to consider the difference between models and patterns. A model is a global description or explanation of a data set, taking a high-level perspective. Models may be descriptive or inferential. Descriptive models seek to summarize the entire data set in a succinct manner. Inferential models aim to provide a mechanism that [...]


