USA Electricity Generating Exploratory Analysis¶

import pandas as pd
import numpy as np
from bokeh.plotting import figure, show, gmap
from bokeh.io import output_file, show, output_notebook, export_png, save, reset_output, export_svg
from bokeh.palettes import brewer
from bokeh.transform import dodge, factor_cmap
from bokeh.models import ColumnDataSource, FactorRange, GMapOptions, HoverTool
output_notebook()

df_2018 = pd.read_csv("raw_data/y2018.csv")

print(df_2018.head(10), "\n")

print(f"Electricity generating technologies in this dataset: {df_2018.Technology.unique()}")

print(f"Number of power plants in this dataset: {df_2018.Plant_ID.nunique()}")

largest_gen_df_2018 = df_2018_clean.sort_values("Electricity Output (MW)", ascending=False) \
                              .drop(columns=["Longitude", "Latitude"])
with pd.option_context('expand_frame_repr', False):
    print(largest_gen_df_2018.head(10))

df_bars = df_2018_clean.copy()
drop_cols = ["Plant_ID", "State", "Longitude", "Latitude"]
# Convert to gigawatts
for c in ["Electricity Output (MW)", "Nameplate"]:
    df_bars[c] = df_bars[c]/10**3

df_bars2 = df_bars.copy()  # For next analysis

tech_group = df_bars.drop(columns=drop_cols) \
                    .groupby("Technology") \
                    .agg({"Electricity Output (MW)": 'sum', "Nameplate": 'sum'}) \
                    .sort_values(by="Nameplate", ascending=False) \
                    .reset_index()

source = ColumnDataSource(tech_group)


p = figure(plot_height=400, x_range=tech_group["Technology"], title="Electricity Output by Technology",
           toolbar_location=None,  y_axis_label = "Gigawatts")
p.vbar(x="Technology", top="Electricity Output (MW)", width=.9, source=source, color="#e6e600")
p.y_range.start = 0
p.xgrid.grid_line_color = None
# output_file("images/output.html")
# save(p)
# export_svg(p, filename="output.svg")
show(p)

df_bars2 = df_bars2.groupby("Technology") \
                   .sum() \
                   .reset_index() \
                   .sort_values("Nameplate", ascending=True)

technologies = list(df_bars2["Technology"])
categories = ["Electricity Output", "Nameplate"]
output = list(df_bars2["Electricity Output (MW)"])
nameplate = list(df_bars2["Nameplate"])

x = [(technology, category) for technology in technologies for category in categories]
counts = sum(zip(output, nameplate), ())
source = ColumnDataSource(data=dict(x=x, counts=counts))


f = figure(y_range=FactorRange(*x), plot_width=800, x_axis_label = "Gigawatts")
f.hbar(y='x', right='counts', width=.9,  color="#e6e600", source=source, line_color="white",
      fill_color=factor_cmap('x', palette=['firebrick', 'olive'], factors=categories, start=1, end=2))
f.ygrid.grid_line_color = None
f.x_range.start = 0
# output_file("images/name_and_output.html")
# save(f)
show(f)

technology = "Solar"
coords = (37.77, -122.42) # San Francisco
plot_technology(df_2018_clean, technology, colors[technology], start_coords=coords)

technology = "Nuclear"
coords = (41.49, -81.69) # Cleveland
plot_technology(df_2018_clean, technology, colors[technology], start_coords=coords)

technology = "Wind"
coords = (41.25, -95.93) # Omaha
plot_technology(df_2018_clean, technology, colors[technology], start_coords=coords)

technology = "Hydroelectric"
coords = (45.5, -122.67) # Portland
plot_technology(df_2018_clean, technology, colors[technology], start_coords=coords)

technology = "Coal"
coords = (40.44, -79.99) # Pittsburgh
plot_technology(df_2018_clean, technology, colors[technology], start_coords=coords)

technology = "Natural Gas"
coords = (33.74, -84.38) # Atlanta
plot_technology(df_2018_clean, technology, colors[technology], start_coords=coords)

technology = "Other"
coords = (32.77, -96.8) # Dallas
plot_technology(df_2018_clean, technology, colors[technology], start_coords=coords)

# Copy and group data. Convert MW to GW
drop_cols = ["Plant_ID", "Latitude", "Longitude", "Technology", "Nameplate"]
df_state = df_2018_clean.copy() \
                        .drop(columns=drop_cols) \
                        .groupby(by=["State", "Classification"]).sum() \
                        .reset_index()
df_state["Electricity Output"] = df_state["Electricity Output (MW)"]/1000

# Pesky Vermont has no fossil fuel generation (I am skeptical about this, but I am not investigating the data that closely)
# Add 0 manually
vermont = pd.Series(data={"State": "VT", "Classification": "Fossil Fuel", "Electricity Output": 0})
df_state = df_state.iloc[:92].append(vermont, ignore_index=True) \
                             .append(df_state.iloc[92:]) \
                             .drop(columns="Electricity Output (MW)") \
                             .reset_index(drop=True)

# Plots
states = list(df_state["State"].unique())
categories = ["Fossil Fuel", "Renewable"]
fossil = list((df_state[df_state["Classification"] == "Fossil Fuel"])["Electricity Output"].round(0))
renew = list((df_state[df_state["Classification"] == "Renewable"])["Electricity Output"].round(0))

x = [(state, category) for state in states for category in categories]
counts = sum(zip(fossil, renew), ())
source = ColumnDataSource(data=dict(x=x, counts=counts))


f = figure(y_range=FactorRange(*x), plot_width=800, plot_height=2000, x_axis_label = "Gigawatts")
f.hbar(y='x', right='counts', width=.9,  color="#e6e600", source=source, line_color="white",
      fill_color=factor_cmap('x', palette=['darkgrey', 'green'], factors=categories, start=1, end=2))
f.ygrid.grid_line_color = None
f.x_range.start = 0

show(f)

years_dfs = {}
output_array = np.zeros([18, 7]) # 18 years, 7 technologies
years = range(2001,2019)
for idx, y in enumerate(years):
    index = str(y)
    # Read files and normalize plant names
    df = pd.read_csv("raw_data/y" + index + ".csv") \
           .rename(columns={"PLNTCODE": "Plant_ID", "PLANT_CODE": "Plant_ID"})

    # Clean dataframe
    df_clean = cleanse_df(df)

    # Convert to Gigawatts for readability 
    df_clean["Electricity Output"] = df_clean["Electricity Output (MW)"]/1000

    # Drop unused and groupby technology. Create % Output for stacked chart
    df_clean = df_clean.drop(columns=["Plant_ID", "Latitude", "Longitude","State", "Electricity Output (MW)", "Nameplate"]) \
                       .groupby(by="Technology").sum() \
                       .reset_index()
    df_clean["% Output"] = 100*df_clean["Electricity Output"] / df_clean["Electricity Output"].sum()

    # Create output % array and years dictionary to store dfs
    output_array[idx,:] = np.array(df_clean["% Output"]).reshape(1,-1)
    years_dfs[index] = df_clean

columns = list(df_clean["Technology"])
df_output_years = pd.DataFrame(data=output_array, columns=columns, index=years)

p = figure(x_range=(2001,2018), y_range=(0,100), title="USA % of Total Electricity by Technology")
p.grid.minor_grid_line_color= '#eeeeee'

p.varea_stack(stackers=columns, x='index', color=brewer["Spectral"][7], legend_label=columns, source=df_output_years)
p.legend.location = "bottom_left"

show(p)