fpetitjean / dba Goto Github PK
View Code? Open in Web Editor NEWDBA: Averaging for Dynamic Time Warping
Home Page: http://www.francois-petitjean.com/Research/
License: GNU General Public License v3.0
DBA: Averaging for Dynamic Time Warping
Home Page: http://www.francois-petitjean.com/Research/
License: GNU General Public License v3.0
When the number of channels is greater than the max length of a series, the squared_DTW failes in the following loop:
for i in range(1, s_len): cost_mat[i, 0] = cost_mat[i-1, 0]+delta_mat[i, 0]
Why should the cost matrix and delta_mat depend on the number of features in my feature vector?
Does your implementation (Python) support DBA for varying time series lengths or does it assume all sequences are of a length T (as mentioned in your paper).
Does your implementation support multivariate time series data?
e.g.
series.shape
(number_of_time_series, length_time_series, dimensionality)
Hi,
I have been playing around with the library by trying to plot the average series. It is clearly visible that the average series is not correct. For any advice I would be greatful.
n_dims = 1
t1=[106.7, 106.7, 99.94, 95.83, 92.84, 90.44, 88.3, 86.12, 82.22, 77.2]
t1_T = np.reshape(t1,(1,-1))
t2= [127.8, 127.8, 122.5, 117.4, 112.2, 107.2, 102.3, 97.47, 92.77, 88.2, 83.79, 79.53, 75.43, 71.49, 67.69, 64.02, 60.42, 56.87, 53.3, 49.6, 45.64, 41.24, 36.27, 30.97, 25.83, 21.09, 16.69, 12.58, 8.671, 4.92, 1.294]
t2_T = np.reshape(t2,(1,-1))
series = [t1_T,t2_T]
plt.figure()
#plotting the synthetic data
plt.plot(range(0,len(t1)), t1 ,label='synthetic data')
plt.plot(range(0,len(t2)), t2 ,label='synthetic data')
plt.legend()
#calculating average series with DBA
average_series = DBA.performDBA(series)
#plotting the average series
for d in range(0,n_dims):
plt.plot(range(0,average_series.shape[1]), average_series[d],label='average data')
plt.legend()
plt.show()
Hi,
I have 10000 time series biological data . each is 12000 in length.
It seems DTW is prohibitively expensive while DBA is very promising.
Can you teach me how to you DBA to speed us calculating distances and cluster all the data?
Thanks a lot.
Huanle
PS: i am a biologist, knowing nothing but a bit of python programming.
First of all thanks a lot for making this code available! I am enjoying using it :)
I aim to create an average of time series that are of a greater length than 1000 (around 9000). I was wondering where this constraint comes from and how I can adjust it. Right now I am slicing my time series per 10, so that the length is less than 1000. However, I really want to test what happens when I use the original time series.
Thanks a lot!
I have around 200 time series, but they are of length 30,000. The DBA algorithm just doesn't converge (as expected). Would there be a way to parallelize the DTW process within your algorithm?
Since I am not from cs background (mechanical engineer), my foundations in time series analysis and coding are a bit weak. However, if you could give me quick walk through on what needs to be done, I suppose I should be able to figure out the solution,
I'm trying to generate multivariate time series synthetic data using DBA. As mentioned in one of the papers cited, there are various methods for weight assigning. How do I add those functions in the code, what needs to be changed? Any help would be much appreciated.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.