Get elements from numpy array only matching datetime

Question

I have a list of tidal height data with a reading every 10 minutes for 1 year that I've loaded in to a list from csv.

The end result I'm trying to achieve is to be able to (line or bar)graph tides and end up with something like this: https://nt.gov.au/__data/assets/pdf_file/0020/430238/2018-mar-tidal-info-ntports-centre-island-graph.pdf

I'm pretty new to programming and have set myself the smaller task of creating a tidal graph from height data for a given day. I would then output multiple graphs to make up a week etc.

 # import numpy as np
 # from datetime import datetime
        DATA:
            010120170010  1.700        
            010120170020  1.650    

    for line in csv_reader:    
        data_times.append(datetime.strptime(line[0], "%d%m%Y%H%M"))
        data_height.append(float(line[2]))

    np_data_times = np.array(data_times)
    np_data_height = np.array(data_height)

create array only with today's heights Is there a better way that does the python equivalent of the SQL 'select * from times where date = today()'? Can I create a dictionary with time: height rather than 2 arrays? (I've read that dicts are unordered so stayed away from that approach)

Plot array divided every 6 hours I'd also like to provide data points to the chart but only show times divided every 3 or 6 hours across the X axis. This would give a smoother and more accurate graph. So far I've only found out how to give data to the x axis and it's labels in a 1:1 fashion when I may want 6:1 or 18:1 etc. Is there a particular method I should be looking at?

# import matplotlib.pyplot as plt
plt.title("Tides for today")
plt.xlabel(datetime.date(real_times[0]))
plt.ylabel("Tide Height")
plt.plot(real_times, real_heights)
plt.show()

Answer 1

Don't use a dictionary. This would make everything slow and hard to handle.
I would suggest to consider using pandas.

Reading in the data would work like this:

import pandas as pd
from datetime import datetime

conv={"Time" : lambda t: datetime.strptime(t, "%d%m%Y%H%M")}
df = pd.read_csv("datafile.txt", header=None, delim_whitespace=True, 
                 names=["Time", "Tide"], converters=conv,index_col=0 )

This results in something like

                     Tide
Time                     
2017-01-01 00:10:00  1.70
2017-01-01 00:20:00  1.65
2017-01-01 05:20:00  1.35
2017-01-02 00:20:00  1.75

You can now filter the dataframe, e.g. for selecing only the data from the first of january:

df["2017-01-01":"2017-01-01"] 

You could directly plot the data like

df["2017-01-01":"2017-01-01"].plot(kind="bar")

or

df["2017-01-01 00:00":"2017-01-01 06:00"].plot(kind="bar")

This will work nicely if the times are equally spaced because it creates a categorical bar plot. (just remember that you might need to use pyplot.show() if working in a script)

You may also use matplotlib to draw the bars

import matplotlib.pyplot as plt
import numpy as np
import matplotlib.dates

df1 = df["2017-01-01 00:00":"2017-01-01 06:00"]
plt.bar(df1.index,df1.Tide, 
        width=np.diff(matplotlib.dates.date2num(df1.index.to_pydatetime()))[0], ec="k")
plt.show()

To get control over the xaxis ticks and labels this latter matplotlib solution would be the way to go. First set the bars to align to the edges, align="edge". Then use formatters and locators as shown in the official dates example. A grid can be defined using plt.grid.

plt.bar(df1.index,df1.Tide, 
        width=np.diff(matplotlib.dates.date2num(df1.index.to_pydatetime()))[0], 
        align="edge", ec="k")

hours = matplotlib.dates.HourLocator()   # every hour
#hours = matplotlib.dates.HourLocator(byhour=range(24)[::3]) # every 3 hours
fmthours=matplotlib.dates.DateFormatter("%m-%d %H:%M")
plt.gca().xaxis.set_major_locator(hours)
plt.gca().xaxis.set_major_formatter(fmthours)

plt.grid(True, axis="x", linewidth=1, color="k")
plt.gcf().autofmt_xdate()
plt.show()

This may then look something like the following: