Hans Skov-Petersen | 11 Mar 16:06 2014
Picon

Pandas: DLL load failed: The specified procedure could not be found

Hi there,

I am trying to make Pandas run with my present Python version.
For a number of reasons I am presently (still) running Python 2.6.
I am on Win7, 64 bit.

I downloaded pandas-0.13.1.win-amd64-py2.6.exe (md5) from https://pypi.python.org/pypi/pandas#downloads

Install is ok, but when importing the Pandas module I get this error:
DLL load failed: The specified procedure could not be found

Any clues?

Cheers
Hans

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org.
For more options, visit https://groups.google.com/d/optout.
Felix Lawrence | 10 Mar 06:15 2014
Picon

Redesigning handling of multidimensional data

Hi,

I've been using pandas for a couple of months now, and I've found it great, but have encountered some awkwardness with indexing multidimensional data. To start a discussion, I've blogged about the problems and have proposed some solutions [1]. The changes I suggest are fairly radical and could transform the way people index multidimensional data in pandas, making people less reliant on group_by et al, and removing the need to stack/unstack.

I don't have the time or technical chops to pull this off by myself in the near future.

Does this vision have any support? Can it be refined and implemented? How do we start?

TLDR: please improve MultiIndexes to the point that MultiIndex + Series is the preferred way to store matrix-style data.

[1] http://camelcode.wordpress.com/2014/02/28/index-to-the-koalas-series-of-posts/

Cheers,
Felix

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org.
For more options, visit https://groups.google.com/d/optout.
Bryan Van de Ven | 10 Mar 19:17 2014

ANN: Bokeh 0.4.2

I am happy to announce the release of Bokeh version 0.4.2!

Bokeh is a Python library for visualizing large and realtime datasets on the web.  Its goal is to provide
elegant, concise construction of novel graphics in the style of Protovis/D3, while delivering
high-performance interactivity to thin clients.  Bokeh includes its own Javascript library (BokehJS)
that implements a reactive scenegraph representation of the plot, and renders efficiently to HTML5
Canvas. Bokeh works well with IPython Notebook, but can generate standalone graphics that embed into
regular HTML.

Check out the full documentation, interactive gallery, and tutorial at

     http://bokeh.pydata.org

If you are using Anaconda, you can install with conda:

     conda install bokeh

Alternatively, you can install with pip:

     pip install bokeh

Some of the new features in this release include:

* Additional Matplotlib and Seaborn compatibility (PolyCollection)
* Extensive tutorial with exercises and solutions added to docs
* new %bokeh magic for improved IPython notebook integration
* Windows support for bokeh-server with two new storage backends (in-memory and shelve)

Also, we've fixed lots of little bugs - see the CHANGELOG for full details.

BokehJS is also available by CDN for use in standalone javascript applications:

     http://cdn.pydata.org/bokeh-0.4.2.js
     http://cdn.pydata.org/bokeh-0.4.2.css
     http://cdn.pydata.org/bokeh-0.4.2.min.js
     http://cdn.pydata.org/bokeh-0.4.2.min.css

Some examples of BokehJS use can be found on the Bokeh JSFiddle page:

     http://jsfiddle.net/user/bokeh/fiddles/

The release of Bokeh 0.5 is planned for late March. Some notable features we plan to include are:

* Abstract Rendering for semantically meaningful downsampling of large datasets
* Better grid-based layout system, using Cassowary.js
* More MPL/Seaborn/ggplot.py compatibility and examples
* Additional tools, improved interactions, and better plot frame
* Touch support

Issues, enhancement requests, and pull requests can be made on the Bokeh Github page: https://github.com/continuumio/bokeh

Questions can be directed to the Bokeh mailing list: bokeh@...

Special thanks to recent contributors: Melissa Gymrek, Amy Troschinetz, Ben Zaitlen, Damian Avila, and
Terry Jones

Regards,

Bryan Van de Ven
Continuum Analytics
http://continuum.io

--

-- 
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe@...
For more options, visit https://groups.google.com/d/optout.

Bill Blount | 8 Mar 20:26 2014
Picon

Plotting TS Data without skips

is this still the case?  seems like a lot of trouble for such a common inconvenience.

from mpl documentation       http://matplotlib.org/examples/pylab_examples/date_index_formatter.html

"""
When plotting daily data, a frequent request is to plot the data
ignoring skips, eg no extra spaces for weekends. This is particularly
common in financial time series, when you may have data for M-F and
not Sat, Sun and you don't want gaps in the x axis. The approach is
to simply use the integer index for the xdata and a custom tick
Formatter to get the appropriate date string for a given index.
"""


thanks in advance.   Bill 

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org.
For more options, visit https://groups.google.com/d/optout.
mail | 7 Mar 18:13 2014
Picon

Free interactive pandas tutorials

Hello everyone,

a friend of mine and I use Python a lot for work and for fun. We also love Wes McKinney's book about Pandas, which proved as a great help learning it.

Unfortunately, there are no interactive tutorials like those that are available at Code School or codecademy for other languages and libraries. Even learnpython.org is "just" about Python itself, not about Pandas. So we decided to do it. We got the technology to implement interactive open source tutorials for Pandas as a web app.

I'm writing this mail to ask whether there is something similar in the making and whether you'd like to tell your thoughts about it. Is there anything that you would like to see? Any feature requests or possible caveats?

We'll use an interactive Python interpreter, simulate some functions and plot everything in the browser using D3. At least, that's the plan. It'll be a long-term project as this seems quite complex.

Looking forward to hearing your thoughts :-)

Best wishes,
Tobias Knuth

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org.
For more options, visit https://groups.google.com/d/optout.
judy wawira | 7 Mar 03:43 2014
Picon

Pandas in Django

Hello 

I have a CSV file that i wish to manipulate based on user inputs and show this on a Djnago project

I dont necessarily wish to save the data in the model 

Can anyone point me to a direction on how to do this 

I have tried creating a normal app like below but it says listmethod not found 

    mydf = pd.read_csv(file,header=1)

    #get header
    cols = list(mydf.columns.values)
    colheader = cols.tolist()

Judy

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org.
For more options, visit https://groups.google.com/d/optout.
Skipper Seabold | 6 Mar 18:14 2014
Picon

UX changes in pandas 0.13.0

The pandas devs wanted to get a bit of feedback on recent UX changes in 0.13.0.

These changes are to the repr of the DataFrame to no longer show the
`info` view by default for large DataFrames. Instead, the DataFrames
now use ... to show that some data has been suppressed and the
addition of a footer showing the number of rows and the columns in the
DataFrame repr, whether truncated or not.

These are big changes to long-standing UX of working with DataFrames,
and we'd be interested to get some user feedback.

The changes were discussed by a few devs here [1] and implemented here [2].

There's some discussion about these changes [3].

Now that that's out there, my concerns. If you read [3], you've seen
them already.

I'd like to see the footer either removed entirely or only displayed
when there's truncation in play. Case for the former, this information
is available already in `info` and in `shape`. Case for the latter, If
it's a small DataFrame, it is unnecessary noise. You can *see* how
many rows and columns there are. Statsmodels uses DataFrames to return
results in certain places. Formerly, this was a nice, tight way to
present information to the user. Now this extra info draws the eye
away from the important information. There's an option to turn this
off, but we can't control what users are doing, so we can't control
what they see and we lose control of our UX a bit.

My less pressing concern is that I'm not wild about the summary
truncation view in general. "We fear change." That's what `head` and
`tail` are for. That said, I can set the option, so it's not that big
of a deal. And the argument is that the summary `info` view was
confusing for new users, which shouldn't be taken lightly.

Skipper

[1] https://github.com/pydata/pandas/issues/4886
[2] https://github.com/pydata/pandas/pull/5550
[3] https://github.com/pydata/pandas/issues/6547

--

-- 
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe@...
For more options, visit https://groups.google.com/groups/opt_out.

Juan Na | 6 Mar 09:58 2014
Picon

Regrouping question

Hi,

I know I'm asking here is easy, and other times I have done, but now I do not know how. I've spent a lot of time to find out and now appeal to you.
I have a dataset (around 89000 records), similar to what I show here. Variables Case, Sta, Zone, Din, Lm, Level, Amin, Amax, Backg can be considered as factors that are assigned values ​​'a priori'.


Case Sta Zone Din Lm Level Amin Amax Backg TLT MTL TTA
1 po1 ip di min ni 1 4 50 1493.21 4.8798 13.854
1 po1 ip di min ni 1 4 100 1515.52 5.5514 15.465
1 po1 ip di min ni 1 4 150 1581.58 6.4032 18.452
1 po1 ip di min ni 1 4 200 1704.93 7.8568 20.563
1 po1 ip di min ni 1 4 250 1823.73 8.2522 26.135
31 pa if nd min ni 2 12 150 1581.58 6.4032 12.652
31 pa if nd min ni 2 12 200 1704.93 7.8568 16.562
32 pa2 if di min nd
1 4 50 1823.73 8.2522 10.256

The thing is I want to know how to perform a series of operations regrouping. Eg. How I can know which is the value of 'Backg' which maximizes the TLT value for each of the groups defined by these variables​​: Case, Sta, Zone, Din, Lm, Level, Amin, Amax, and what is this value of TLT, ...
 

I think I use groupby, so that.:

df = pd.read_csv ("mifile.LOG")
dfg =df.groupby (.....)
And then?


Sorry for include this newbie question, but I need a track because I'm a little lost

Thanks
Juan

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org.
For more options, visit https://groups.google.com/groups/opt_out.
ASHISH KUMAR GUPTA | 5 Mar 09:02 2014
Picon

SettingWithCopyWarning

Hi,

I have noticed that when I try to do shorting or filling with "inplace=True" option in pandas.DataFrame object, it gives me warning SettingWithCopyWarning. The weird things is that sometime rerunning the same piece of code doesn't give the warning. Can someone please explain what this means and how I can avoid such warnings?

Thanks
Ashish

SettingWithCopyWarning: A value is trying to be set on a copy of a slice from a DataFrame
df.fillna(0.0, inplace = True)



--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org.
For more options, visit https://groups.google.com/groups/opt_out.
Peter Prettenhofer | 28 Feb 09:39 2014
Picon

Binary columns and missing values

Hi all,

I've a csv file with lots of binary columns. True values are encoded as 'true' and missing values indicate False values. When I use ``pandas.read_csv('file.csv', true_values=['true'])`` the columns have dtype ``object`` because of the missing values. When I properly encode False values as 'false' then the columns are indeed bool. Is there a way to treat columns with one value and missing values as boolean? It would be a huge memory safer for me - the proper encoded version has a 4x larger file size but 10x smaller memory footprint as a DataFrame.

thanks,
 Peter

PS: if its not possible - is that something people are interested in and if so should I prepare a PR for it?

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
For more options, visit https://groups.google.com/groups/opt_out.
Stephan Hoyer | 25 Feb 03:41 2014
Picon

Why does pandas.Float64Index have dtype=object?

I am working on a package [1] that uses pandas.Index objects internally to allow for fast label based indexing of sets of n-dimensional arrays.

One quirk we're noticed is that pandas.Float64Index objects have dtype=object. This means that mathematical operations on indices (e.g., 2 * index) will always return an array with dtype=object, even if I do something like 2 * np.asarray(index). For us, this is annoying, because object arrays are slow.

From our perspective, it would be great if pandas.Index objects, when created from a numpy.ndarray, maintained the dtype of the original array -- or, at the very least, if they exposed their array interface with the original data type [2]. Presumably, there is a reason why this isn't the case? Perhaps it's something to consider addressing in the refactor of Index for pandas 0.14?

We would be very curious to understand this behavior.

Cheers,
Stephan

[2] I suppose we will probably create our own wrapper of pandas.Index objects that modifies __array__ in this fashion.

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
For more options, visit https://groups.google.com/groups/opt_out.

Gmane