Working on it, Made a few graphs

BekksRoom/messy.py

@@ -0,0 +1,139 @@
+from BekksRoom.stats_functions import average_minutes, stdev_minutes
+from BekksRoom.stats_functions import hypothesis_test_plot
+import pandas as pd
+import matplotlib as plt
+
+
+def get_input(dataset):
+    done = False
+    std = 'n'
+    avg = 'n'
+    hypothesis = 'n'
+    going_to_use_groupby = 'n'
+    avg = input("want the average? y/n ")
+    clean = input("Want your data cleaned?  We drop zeros and nan. y/n ")
+    columns = []
+    if avg == "y":
+        print("Automatically including weight because you want the average")
+        columns.append('weight')
+    else:
+        std = input("Want standard Deviation? y/n ")
+        if std == 'y':
+            print("Automatically including weight because you want the standard dev")
+            columns.append('weight')
+    if avg != 'y' or std != 'y':
+        hypothesis = input("Want hypothesis testing? y/n ")
+        if hypothesis == 'y':
+            print("Automatically including weight because you want hypothesis test")
+            columns.append('weight')
+    if avg != 'y' and std != 'y' and hypothesis != 'y':
+        going_to_use_groupby = input("Are you going to use groupby? y/n ")
+        if going_to_use_groupby == 'y':
+            print("Automatically including weight because you want to use groupby ")
+            columns.append('weight')
+    for item in dataset.columns.tolist():
+        print(item)
+    while not done:
+        print("Enter you column as appears above, type DONE when done, or CLEAR if you messup")
+        column = input("What columns do you want?")
+        if column == 'DONE':
+            done = True
+            break
+        if column == 'CLEAR':
+            columns = []
+        if column in dataset.columns.tolist():
+            columns.append(column)
+        else:
+            print("Not in dataset")
+        print(columns)
+    return(avg, clean, columns, std, hypothesis, going_to_use_groupby)
+
+
+def querys(datawork, query):
+    return datawork[query]
+
+
+def graph_data(dataset):
+    wplot = input("What type of plot? bar or line (more to come)")
+    if wplot == "bar":
+        print(plt.style.available)
+        plot_style = input("What style plot? Type it correctly!")
+        plt.style.use(plot_style)
+        print("Your data sir ", dataset)
+        figuresize = input("Whats your figure size, just enter one number no floats")
+        figuresize = int(figuresize)
+        legends = input("Want a legend True/False")
+        stacks = input("Want to stack it? True/False")
+        xlabels = input("x label?")
+        ylabels = input("y label?")
+        grids = input("Want a grid True/False")
+        ax = dataset.plot(kind="bar", figsize=(figuresize,figuresize), legend=eval(legends), stacked=eval(stacks))
+        ax.set_ylabel(ylabels)
+        ax.grid(eval(grids), 'both')
+        ax.set_xlabel(xlabels)
+        return ax
+    if wplot == "line":
+        print(plt.style.available)
+        plot_style = input("What style plot? Type it correctly!")
+        plt.style.use(plot_style)
+        print("Your data sir ", dataset)
+        figuresize = input("Whats your figure size, just enter one number no floats")
+        figuresize = int(figuresize)
+        legends = input("Want a legend True/False")
+        xlabels = input("x label?")
+        ylabels = input("y label?")
+        grids = input("Want a grid True/False")
+        ax = dataset.plot(kind="line", figsize=(figuresize,figuresize), legend=eval(legends))
+        ax.set_ylabel(ylabels)
+        ax.grid(eval(grids), 'both')
+        ax.set_xlabel(xlabels)
+        return ax
+
+
+def datasets(dataset, query=None):
+    avg, clean, columns, std, hypothesis, going_to_use_groupby = get_input(dataset)
+    # print(avg, clean, columns)
+    data = dataset[columns]
+    if query is not None:
+        data = querys(data, query)
+    data.describe()
+    if std == 'y':
+        newdataframe = {}
+        for item in data.columns.tolist():
+            if item == "weight":
+                pass
+            else:
+                newdataframe[item] = [stdev_minutes(data, item)]
+        data = pd.DataFrame.from_dict(newdataframe, orient='columns', dtype=None)
+    if avg == 'y':
+        newdataframe = {}
+        for item in data.columns.tolist():
+            if item == "weight":
+                pass
+            else:
+                newdataframe[item] = [average_minutes(data, item)]
+        data = pd.DataFrame.from_dict(newdataframe, orient='columns', dtype=None)
+    if clean == 'y':
+        for item in data.columns.tolist():
+            if item == "weight":
+                pass
+            else:
+                data = data[data[item] > 0]
+                data = data.dropna()
+                # print(data)
+    if going_to_use_groupby == 'y':
+        while True:
+            column_to_group = input("What column would you like to group by? ")
+            if column_to_group in data:
+                return data.groupby(column_to_group)
+            else:
+                print("Thats not in your columns")
+    if hypothesis == 'y':
+        group_var = input("What is your group variable?")
+        test_var = input("What is your test variable?")
+        data = hypothesis_test_plot(data, group_var, test_var)
+    graphs = input("Want to graph your data? y/n")
+    if graphs == 'y':
+        plot = graph_data(data)
+        return(data, plot)
+    return data

BekksRoom/stats_functions.py

@@ -0,0 +1,38 @@
+import math
+import pandas as pd
+
+
+def average_minutes(data, activity_col, weight_col="weight"):
+    data = data.rename(columns={weight_col: "weight",
+                                activity_col: "minutes"})
+    data = data[['weight', "minutes"]]
+    data['weighted_minutes'] = data.weight * data.minutes
+    return data.weighted_minutes.sum() / data.weight.sum()
+
+
+def stdev_minutes(data, activity_col, weight_col="weight"):
+    data_mean = average_minutes(data, activity_col, weight_col)
+    data = data.rename(columns={weight_col: "weight",
+                                activity_col: "minutes"})
+    num_non0_obs = data[data.weight != 0].weight.count()
+    data = data[["weight", "minutes"]]
+    data['weighted_ss'] = data.weight * (data.minutes - data_mean)**2
+    return math.sqrt(data.weighted_ss.sum() /
+                     (((num_non0_obs-1)/num_non0_obs)*data.weight.sum()))
+
+
+def hypothesis_test_plot(data, group_var, test_var, weight_var="weight"):
+    data = data[[group_var, test_var, weight_var]]
+    data_grouped = data.groupby(group_var)
+    frame = pd.DataFrame()
+    for group in data_grouped:
+        count = group[1][weight_var].count()
+        mean = average_minutes(group[1], test_var, weight_var)
+        stdev = stdev_minutes(group[1], test_var, weight_var)
+        frame = frame.append({group_var: group[0],
+                              "mean": mean,
+                              "error": (stdev*1.96/math.sqrt(count))},
+                             ignore_index=True)
+    frame.index = frame.pop(group_var)
+    plot = frame.plot(kind="bar", yerr="error", figsize=(12, 8))
+    return (frame, plot)