Hi 

From my understanding  this likely means that “trend” offers no predictive 
quality to the dataset. Meaning that anomaly scores or CLA predictions were not 
improved, or diminished,when adding the “trend” data item. 

I’ve had the same experience in my own testing. Data items I thought should 
have a positive influence on prediction often had the reverse affect. 

-Phil



> On Jan 5, 2016, at 5:24 AM, Raf <[email protected]> wrote:
> 
> Sorry in my last mail I meant NuPIC instead of "Numenta"!
> Cheers.
> 
> On 05/01/2016 11:55, Michael Lam wrote:
>> Hi,
>> 
>> I have been trying to use multiple parameters based on the hot gym tutorial 
>> with some modifications, such as added an extra included field:
>> 
>> SWARM_DESCRIPTION = {
>>  "includedFields": [
>>    {
>>      "fieldName": "timestamp",
>>      "fieldType": "datetime"
>>    },
>>    {
>>      "fieldName": "price",
>>      "fieldType": "float",
>>      "maxValue": 1000.0,
>>      "minValue": 0.0
>>    },
>>    {
>>      "fieldName": "trend",
>>      "fieldType": "float",
>>      "maxValue": 20.0,
>>      "minValue": -20.0
>>    }
>>  ],
>> 
>> notice I added "trend" in addition to "price", but after swarming, the 
>> resulting params.csv file only has time and price in it, no trend.  Can 
>> someone tell me what I am doing wrong?  Thanks.
> 
> -- 
> Raf
> 
> www.madraf.com/algotrading
> reply to: [email protected]
> skype: algotrading_madraf
> 
> 
> 


Reply via email to