C3_W1_Anomaly_Detection_ the array p_val

Course Q&A Machine Learning Specialization Unsupervised Learning, Recommenders, Reinforcement
1

The array p_val given by multivariate_gaussian(X_val, mu, var) is always only zeros.
mu and var seem fine, and also the X_val values are close to mu.
Shouldnt the p_val colose to mu be higher?

2

Hello @David_Gschliesser,

Did you mean exactly zero or close to zero?

Can you give 2 to 3 examples such as what X_val, mu, var you used and what p_val you got?

Raymond

3

Thanks for replying!
Sadly exactly zero, when I write p_val[p_val==0]=2 i get a all 2 array.
my arrays are:
mu=[14.11222578 14.99771051]
var=[1.83263141 1.70974533]
and (sorry long but i want to show the last line) X_val=
[[15.79025979 14.9210243 ]
[13.63961877 15.32995521]
[14.86589943 16.47386514]
[13.58467605 13.98930611]
[13.46404167 15.63533011]
[12.94888838 16.14006828]
[15.31084155 15.17480137]
[13.89795241 15.43169469]
[11.92096525 14.30579937]
[14.85934119 14.90066474]
[13.63501268 14.5620583 ]
[15.75981741 15.13340409]
[15.221965 14.18491044]
[12.10564111 13.42653822]
[14.80848889 13.87075725]
[15.36135887 15.99014917]
[12.72914442 15.8265107 ]
[15.38533789 15.21036416]
[13.81810534 14.29933915]
[15.3788488 15.90228514]
[14.11232153 15.31616729]

i get all zeros, even for the last one

But if i give only

If i give as values:
mu=[14.11222578, 14.99771051]
var=[1.83263141, 1.70974533]
X_val=[[14.11232153, 15.31616729]]

i get p_val=[0.0872844]

the shapes are:
mu.shape is (2,)
var.shape is (2,)
X_val.shape is (307,2)

thanks for your help, somehow i am very stuck… but i have the feeling its just a little thing