4

Click here to load reader

LOGIC SYSTEM ASSIGNMENT.pdf

Embed Size (px)

Citation preview

  • LOGIC SYSTEM ASSIGNMENT

    ANUP KUMAR SAH ROLL NO- 70282

    SEMSESTER VI

    DATA SET IRIS DATA SET 50 Instances , 4 attributes

    SOURCE: Creator: R.A. Fisher Donor: Michael Marshall (MARSHALL%PLU '@' io.arc.nasa.gov)

    ATTRIBUTE INFORMATION 1. sepal length in cm 2. sepal width in cm 3. petal length in cm 4. petal width in cm

    5. class: Iris Setosa represented with value 1 Iris Versicolour represented with value 2 iris Virginica represented with value 3

  • Regression Problem Predict real-valued output

    M=Number of training examples X= represents input features Y= Represent output target value

    1. Missing values in data There was error in the fourth feature of 35th sample and error in error were in second and third feature of 38thsample. The problem does not have any missing feature, in case of missing attribute we can take mean of that feature attribute and replace with that value.

    2. Attribute Selection

    We have four different features for sepal length in cm, sepal width, sepal length in cm, petal length in cm, petal width in cm. I have taken all the features in our regression model Feature selection techniques provide three main benefits when constructing predictive models: improved model interpretability low training set But for our prediction I have taken all the features of 150 data set for our regression model.. 3. Approach to build Prediction Model

    Features X Estimated of data class

    Training Set

    Learning Algorithm

    h (hypothesis)

  • Hypothesis ho() = o + 1 Cost Function

    J()= 12

    (ho()=1 ())2 Our goal is to minimize cost.

    X0

    X1 X2 X3 X4 Y

    1 5.1 3.5 1.4 .2 1 1 4.9 3 1.4 .2 1 1 4.7 3.2 1.3 .2 1 . . . . . . . . . . . . . . . . . .

    1 5.9 3.0 5.1 1 3 Normal Equation Method And Theta is being calculated by normal Equation , method. = ()1 4. Analysis of Validation Model After the regression coefficient has been obtained. We have a predictor which takes a feature vector and returns an output. To validate the model, we have checked the prediction of our model for each known instances which is in strong admittance with the actual value as prescribed in data set. Since the prediction output has only 3 possible values {1, 2, 3}. We might consider the rounding-off of predicted values which makes all the predictions to almost uniformly converge to the actual value.