01. package: matplotlib.pyplot, method: plot, data: two python lists

test setup
• install python 3, pandas, ... with anaconda3
• mac, code visual studio
• create a folder, 01_lines
• add a file, test.py

       import matplotlib.pyplot
       plt.plot([1,2,3,4],[2,8,5,9], label="test 1") 
       plt.plot([1,2,3,4],[9,2,7,3], label="test 2")

       plt.xlabel("x")
       plt.ylabel("y")
       plt.title("plot method demo\n12/22/2018")
       plt.legend()

       plt.savefig("01_lines/two_lines.png")
            

code review
• The method draws lines
• The first list is the x-coordinates for 4 points.
• The second list is the y-coordinates for 4 points.
• The method builds the graphics in background.
• Method savefig creates a image file for rendering.
• No pandas, no dataframe is involved

01x. one pandas dataframe

introduction
• The previous topic uses hard coded data.
• Using csv files for data is more realistic.
• Module pandas provides dataframe type.
• Both module python and pandas are built upon module numpy.
• Using numpy, the efficiency is high. Because there is no label-like variables, very static.

steps
• in code visual studio
• Under the working folder, 01_lines, create a file, xy32.txt as below
     1,2
     2,8
     3,5
  
• One row stands for one point - x, y.
• create a python file as below

        import pandas as pd
        data = pd.read_csv('01_lines/xy32.txt', names =["x", "y"])
        v = data.values
        t = v.transpose()          #[[1 2 3], [2 8 5]]
        x = t[0]                   #[1 2 3]
        y = t[1]                   #[2 8 5]
              
        import matplotlib.pyplot as plt
        plt.plot(x,y)
        plt.savefig("01_lines/lines11.png")
            

code analysis
• create a csv file from a txt file
• Argument names for columns are needed, otherwise, the 1st row will be used for column names
• v = data.values removes column names and row names.
• v.transpose() swaps rows and columns.

Other approaches
• In the text file, one row for the x-coordinates for these three points.
• The other row for the y-coordinates for these three points.
• There are two rows only. No transpose is needed
• The text file can be stored in a remote web site.
  It can be accessed with HTTP protocal.
  Use the same method, like pd.read_csv('http://....',...)

one scenario example to use plot method
• There are two column variables involved.
• A chicken growing scenario is as below:
• The data for x can be months.
• The data for y can be the weight sum of a group of chicken.
• Lines are drawn between days continuously.
• That's the way to monitor the chicken weight growth.
• Two tests are for different foods.

02. bar chart

• the senario to use bar.
• stock prices in y-coordinate
• stock names in x-coordinate
• the bar in green for its start price.
• the bar in green for its end price.
• In x, stocks are discrete, not continuous.
• Six lists must be prepared from some data source like csv for use.

code as below:

        print("-----------   bar.py  --------------------")
        import matplotlib.pyplot as plt

        fig = plt.figure(figsize=(7,5))

        names = ["stock A", "stock B", "stock C", "stock D"]
        price1 = [200, 390, 450, 175]
        price2 = [180, 310, 350, 220]
        positions1 = [0,    1,    2,    3]
        positions2 = [0.3,  1.3,  2.3,  3.3]
        positions3 = [0.15, 1.15, 2.15, 3.15]

        plt.bar(positions1,  price1,  width=0.3, color="g", label="price 1")
        plt.bar(positions2,  price2,  width=0.3, color="r", label="price 2")

        plt.xticks(positions3, names)
        plt.legend()

        plt.savefig("02_bar/bar.png")

        print("---------  end of test 1 -------")
            

03. Histogram

• A histogram graphics LOOKS LIKE a bar graphics. THEY ARE VERY DIFFERENT ANIMALS.
• In this example, there are 80 weight data
• Based on the data, create bins for different age range, like 0-20, 20-40,.. 180-200.
• Use hist method to create a distribution graphics.
• In x, weight ranges are continuous, not discrete.

code as below:

        print("-----------   hist.py  --------------------")
        import matplotlib.pyplot as plt
        fig = plt.figure(figsize=(7,5))

        # prepare data from simulation, 80 weight data from 0 to 200
        import random
        all_weight = []
        for i in range(80):
            w = random.randint(0,200)
            all_weight.append(w)
        print(all_weight)

        # create bins based on the data
        bins = [0,20,40,60,80,100,120,140,160,180,200]

        plt.hist(all_weight, bins, histtype='bar', rwidth=0.7)

        plt.savefig("03_histogram/hist.png")     # see the weight distribution
        print("---------  end of test 4 -------
        
            

04. Scatter Graphics

• You'll see many dot dot dot
• This is the style for determine the correlation between two variables.
• There are three types of correlations
  • Positive: as one variable increases so does the other.
  • Negative Correlation: as one variable increases, the other decreases.
  • none

code as below:

        print("-----------   scatter.py  --------------------")
        import matplotlib.pyplot as plt
        fig = plt.figure(figsize=(7,5))

        x = [1,2,3,4,5,6,7,8,9]
        y = [3,5,5,6,7,9,7,12,15]
        plt.scatter(x, y, color='g', s=100, marker="o")

        plt.savefig("04_scatter/scatter.png")    
        print("---------  end of test 2 -------")
        
            

05. stackplot graphics

• In this scenario, the group is weekdays.
• It has 7 elements.
• Each day element has 4 parts - working, tv, eating, sleep.
• You can see the time usages.

code as below:

        print("-----------   05_stackplot.py  --------------------")
        import matplotlib.pyplot as plt
        fig = plt.figure(figsize=(7,5))

        # group data for x-coordinate
        weekdays = [1, 2, 3, 4, 5, 6, 7]      # x-coordinate

        # parts data for y-coordinate
        working  = [8, 7, 6, 9, 6, 8, 8]      # part 1
        tv       = [2, 3, 3, 2, 3, 0, 5]      # part 2
        eating   = [3, 1, 2, 3, 2, 2, 1]      # part 3
        sleep    = [10,9, 7, 9,11, 10,8]      # part 4

        # for legend
        plt.plot([],[],color='g', label='working', linewidth=5)
        plt.plot([],[],color='c', label='tv',      linewidth=5)
        plt.plot([],[],color='r', label='eating',  linewidth=5)
        plt.plot([],[],color='b', label='sleep',   linewidth=5)

        # creating graphics
        plt.stackplot(weekdays, working, tv, eating, sleep, colors=('g','c','r','b'))
        plt.legend()

        # rendering graphics
        plt.show()  
        #plt.savefig("stackplot.png")      
        print("---------  end of test 2 -------")
        
            

06. pie graphics

• This scenario is about the cost analysis of a security projects.
• One list data is for all the cost elements.
• Another list is for their cost weights.
• labels are for cost element names.

code as below:

        import matplotlib.pyplot as plt
        fig = plt.figure(figsize=(7,5))

        items   = ['parts', 'labor', 'design', 'management', 'misc']
        weights = [100,     80,      50,       40,           60    ]
        cols    = ['y',     'c',     'r',      'g',          'm']

        plt.pie(weights, labels=items,
                colors=cols, startangle=90, autopct='%1.1f%%')
        plt.title('security project cost analysis')

        plt.show()