Skip to content

Commit

Permalink
cosmetic changes to hello world
Browse files Browse the repository at this point in the history
  • Loading branch information
@hosseinmoein
hosseinmoein committed Mar 31, 2024
1 parent 102a353 commit bd918e4
Showing 1 changed file with 70 additions and 94 deletions.
164 changes: 70 additions & 94 deletions examples/hello_world.cc
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,27 +2,21 @@
Copyright (c) 2019-2026, Hossein Moein
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of Hossein Moein and/or the DataFrame nor the
names of its contributors may be used to endorse or promote products
derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL Hossein Moein BE LIABLE FOR ANY
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following
conditions are met:
* Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following
disclaimer in the documentation and/or other materials provided with the distribution.
* Neither the name of Hossein Moein and/or the DataFrame nor the names of its contributors may be used to endorse or promote
products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING,
BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
EVENT SHALL Hossein Moein BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
DAMAGE.
*/

#include <DataFrame/DataFrame.h> // Main DataFrame header
Expand Down Expand Up @@ -69,14 +63,13 @@ struct MyData {
//
int main(int, char *[]) {

// If you want to fully take advantage of DataFrame parallel computing logic,
// it is recommended to call the following at the beginning of your program.
// If you want to fully take advantage of DataFrame parallel computing logic, it is recommended to call the following
// at the beginning of your program.
//
// NOTE: make sure you read and understand the Multithreading section
// in the documentations (threads could potentially hinder performance).
// This program (hello world) is a perfect example. Since I know this program doesn’t
// deal with large datasets to trigger multithreaded algorithms, populating the thread-pool
// with threads (i.e. calling set_optimum_thread_level()) is a waste of resources.
// NOTE: make sure you read and understand the Multithreading section in the documentations (threads could
// potentially hinder performance). This program (hello world) is a perfect example. Since I know this program
// doesn’t deal with large datasets to trigger multithreaded algorithms, populating the thread-pool with threads
// (i.e. calling set_optimum_thread_level()) is a waste of resources.
//
ThreadGranularity::set_optimum_thread_level();

Expand All @@ -87,12 +80,11 @@ int main(int, char *[]) {

ULDataFrame ul_df1;

// One way to load data into the DataFrame is one column at a time.
// A DataFrame column could be at most as long as its index column. So, you must load the indexfirst before
// loading any column.
// One way to load data into the DataFrame is one column at a time. A DataFrame column could be at most as long as its
// index column. So, you must load the indexfirst before loading any column.
//
// Once you load a column or index, the data is moved to DataFrame. The original vectors are now empty.
// There are other ways of loading data without the move.
// Once you load a column or index, the data is moved to DataFrame. The original vectors are now empty. There are other
// ways of loading data without the move.
//
ul_df1.load_index(std::move(idx_col1));
ul_df1.load_column("dbl_col", std::move(dbl_col1));
Expand All @@ -107,8 +99,7 @@ int main(int, char *[]) {

ULDataFrame ul_df2;

// Also, you can load data into a DataFrame all at once. In this case again the data is moved to the
// DataFrame.
// Also, you can load data into a DataFrame all at once. In this case again the data is moved to the DataFrame.
//
ul_df2.load_data(std::move(idx_col2),
std::make_pair("string col", str_col),
Expand All @@ -117,18 +108,17 @@ int main(int, char *[]) {

StrDataFrame ibm_df;

// Also, you can load data into a DataFrame from a file, supporting a few different formats.
// If the file cannot be found, an exception will be thrown.
// If the DataFrame root directory is your current directory when running this, it should work fine.
// Also, you can load data into a DataFrame from a file, supporting a few different formats. If the file cannot be found,
// an exception will be thrown. If the DataFrame root directory is your current directory when running this, it should
// work fine.
//
ibm_df.read("data/IBM.csv", io_format::csv2);

// To access a column, you must know its name (or index) and its type.
// In case of a "standard" DataFrame (not a view), the columns are returned as a reference to a
// std::vector of type of that column.
// To access a column, you must know its name (or index) and its type. In case of a "standard" DataFrame (not a view),
// the columns are returned as a reference to a std::vector of type of that column.
//
// get_column() involves 1 or sometimes 2 hash-table lookups.
// So, you should not call it repeatedly in a loop. Instead get a reference to it and use the reference.
// get_column() involves 1 or sometimes 2 hash-table lookups. So, you should not call it repeatedly in a loop. Instead
// get a reference to it and use the reference.
//
const auto &cool_col_ref = ul_df2.get_column<std::string>("Cool Column");
const auto &str_col_ref = ul_df2.get_column<std::string>("string col");
Expand All @@ -139,27 +129,24 @@ int main(int, char *[]) {
std::cout << str << ", ";
std::cout << std::endl;

std::cout << "There are " << ibm_df.get_column<double>("IBM_Close").size()
<< " IBM close prices" << std::endl;
std::cout << "There are " << ibm_df.get_column<double>("IBM_Close").size() << " IBM close prices" << std::endl;
std::cout << "There are " << ibm_df.get_index().size() << " IBM indices" << std::endl;

// You can write the data to a file or stdout in a few formats. You must specify all the column types,
// but only once. When writing to a file, the file name/path must be create-able.
// You can write the data to a file or stdout in a few formats. You must specify all the column types, but only once.
// When writing to a file, the file name/path must be create-able.
//
ul_df2.write<std::ostream, std::string, double>(std::cout, io_format::csv2);
ibm_df.write<double, long>("/tmp/test.json", io_format::json);

// You can convert a DataFrame to a string and from a string back into a DataFrame.
// This could be used to transmit a DataFrame from one place to another or store a DataFrame in
// databases, caches, …
// You can convert a DataFrame to a string and from a string back into a DataFrame. This could be used to transmit a
// DataFrame from one place to another or store a DataFrame in databases, caches, ...
//
const std::string ibm_df_as_str = ibm_df.to_string<double, long>();
StrDataFrame ibm_df_2;

// Since we convert from native type to string and back, if you have floating point numbers with
// long precisions, you may run into precision mismatches.
// to_string() has a precision parameter you can adjust. The default is 12 which is a relatively
// high precision.
// Since we convert from native type to string and back, if you have floating point numbers with long precisions, you may
// run into precision mismatches. to_string() has a precision parameter you can adjust. The default is 12 which is a
// relatively high precision.
//
ibm_df_2.from_string(ibm_df_as_str.c_str());
// std::cout << ibm_df_as_str << std::endl; // Large output
Expand All @@ -176,8 +163,7 @@ int main(int, char *[]) {
// You must specify all the column types, but only once.
//
auto above_150_fun = [](const std::string &, const double &val)-> bool { return (val > 150.0); };
auto above_150_df =
ibm_df.get_data_by_sel<double, decltype(above_150_fun), double, long>("IBM_Close", above_150_fun);
auto above_150_df = ibm_df.get_data_by_sel<double, decltype(above_150_fun), double, long>("IBM_Close", above_150_fun);

// Or, you could choose to get a view. See docs for views.
//
Expand All @@ -187,58 +173,52 @@ int main(int, char *[]) {
// You can get another DataFrame by group-bying on one or multiple columns.
// You must specify only the type(s) of column(s), you are group-bying.
//
// Group-by column dbl_col, and I am specifying how to summarize the index column and each of the
// other columns.
// Group-by column dbl_col, and I am specifying how to summarize the index column and each of the other columns.
//
auto gb_df =
ul_df1.groupby1<double>("dbl_col",
LastVisitor<ul_idx_t, ul_idx_t>(),
std::make_tuple("integers", "sum_int", SumVisitor<int>()),
std::make_tuple("my_data_col", "last_my_data", LastVisitor<MyData>()));
auto gb_df = ul_df1.groupby1<double>("dbl_col",
LastVisitor<ul_idx_t, ul_idx_t>(),
std::make_tuple("integers", "sum_int", SumVisitor<int>()),
std::make_tuple("my_data_col", "last_my_data", LastVisitor<MyData>()));

// You can run statistical, financial, ML, … algorithms on one or multiple columns by using visitors.
// You must specify the column(s) type(s).
// The visitor's data column is of type double and its index column is of type std::string.
// You can run statistical, financial, ML, … algorithms on one or multiple columns by using visitors. You must specify
// the column(s) type(s). The visitor's data column is of type double and its index column is of type std::string.
//
StdVisitor<double, std::string> stdev_v;

ibm_df.visit<double>("IBM_Close", stdev_v);
std::cout << "Standard deviation of IBM close prices: " << stdev_v.get_result() << std::endl;

// Now let’s declare two DataFrames with index type of DateTime which is a handy object for
// date/time manipulations.
// Now let’s declare two DataFrames with index type of DateTime which is a handy object for date/time manipulations.
//
DTDataFrame ibm_dt_df;
DTDataFrame aapl_dt_df;

// Let’s read the AAPL and IBM market data from their files. The data for these two stocks start and end
// at different dates. But there is overlapping data between them.
// Let’s read the AAPL and IBM market data from their files. The data for these two stocks start and end at different
// dates. But there is overlapping data between them.
//
ibm_dt_df.read("data/DT_IBM.csv", io_format::csv2);
aapl_dt_df.read("data/DT_AAPL.csv", io_format::csv2);

// First let’s make sure if there are missing data in our important columns, we fill them up.
//
ibm_dt_df.fill_missing<double>({ "IBM_Close", "IBM_Open", "IBM_High", "IBM_Low" },
fill_policy::linear_interpolate);
ibm_dt_df.fill_missing<double>({ "IBM_Close", "IBM_Open", "IBM_High", "IBM_Low" }, fill_policy::linear_interpolate);

// Now we join the AAPL and IBM DataFrames using their indices and applying inner-join policy.
//
DTDataFrame aapl_ibm =
ibm_dt_df.join_by_index<DTDataFrame, double, long>(aapl_dt_df, join_policy::inner_join);
DTDataFrame aapl_ibm = ibm_dt_df.join_by_index<DTDataFrame, double, long>(aapl_dt_df, join_policy::inner_join);

// Now we calculate the Pearson correlation coefficient between AAPL and IBM close prices.
// The visitor's data columns are of type double and its index column is of type DateTime.
// Now we calculate the Pearson correlation coefficient between AAPL and IBM close prices. The visitor's data columns are
// of type double and its index column is of type DateTime.
//
CorrVisitor<double, DateTime> corrl_v;

std::cout << "Correlation between AAPL and IBM close prices: "
<< aapl_ibm.visit<double, double>("AAPL_Close", "IBM_Close", corrl_v).get_result()
<< std::endl;

// Now let’s do something more sophisticated and calculate rolling exponentially weighted correlations
// between IBM and Apple close prices. Since this is a rolling -- moving -- analysis the result is a
// vector of exponentially weighted correlations for each date in the data stream.
// Now let’s do something more sophisticated and calculate rolling exponentially weighted correlations between IBM and
// Apple close prices. Since this is a rolling -- moving -- analysis the result is a vector of exponentially weighted
// correlations for each date in the data stream.
//
ewm_corr_v<double> ewmcorr { exponential_decay_spec::span, 3 };
const auto &ewmcorr_result =
Expand All @@ -264,8 +244,7 @@ int main(int, char *[]) {
std::make_tuple("AAPL_Close", "Median", MedianVisitor<double, dt_idx_t>()),
std::make_tuple("AAPL_Close", "25% Quantile", QuantileVisitor<double, dt_idx_t>(0.25)),
std::make_tuple("AAPL_Close", "Std", StdVisitor<double, dt_idx_t>()),
// "Mode" column is a column of std::array<ModeVisitor::DataItem, 2>'s -- It cannot be printed
// by default
// "Mode" column is a column of std::array<ModeVisitor::DataItem, 2>'s
std::make_tuple("AAPL_Close", "Mode", ModeVisitor<2, double, dt_idx_t>()),
std::make_tuple("AAPL_Close", "MAD", MADVisitor<double, dt_idx_t>(mad_type::mean_abs_dev_around_mean)),
// "Z Score" column is a column of std::vector<double>'s
Expand All @@ -285,11 +264,9 @@ int main(int, char *[]) {
// ---------------------------------------------------
//
// Now let’s do some stuff that are a little more involved (multi steps). There are a lot of theories,
// math, and procedures that I am skipping to explain here.
// See docs for more details.
// math, and procedures that I am skipping to explain here. See docs for more details.
//
// NOTE: I am applying the following analysis to financial data but it equally applies to other
// scientific fields.
// NOTE: I am applying the following analysis to financial data but it equally applies to other scientific fields.
//
// ---------------------------------------------------

Expand Down Expand Up @@ -339,19 +316,18 @@ int main(int, char *[]) {
<< "75% quantile: " << qt75.get_result() << ", "
<< "95% quantile: " << qt95.get_result() << std::endl;

// Now let’s do another interesting thing. Let’s take the IBM returns curve and split it into
// 3 different curves; Trend, Seasonal, and Idiocentric or Residual or Random.
// For the sake of this exercise, we assume IBM business goes through 170-day seasonal cycles.
// Now let’s do another interesting thing. Let’s take the IBM returns curve and split it into 3 different curves; Trend,
// Seasonal, and Idiocentric or Residual or Random. For the sake of this exercise, we assume IBM business goes through
// 170-day seasonal cycles.
//
DecomposeVisitor<double, DateTime> decom { 170, 0.6, 0.01 };

// After this call, the 3 curves will be in decom visitor instance. See docs how to get them
// and analyze them.
// After this call, the 3 curves will be in decom visitor instance. See docs how to get them and analyze them.
//
ibm_dt_df.single_act_visit<double>("IBM_Return", decom);

// But what if you don’t know the seasonality of IBM returns which would be most of the time.
// No worries, Mr Joseph Fourier comes to the rescue.
// But what if you don’t know the seasonality of IBM returns which would be most of the time. No worries,
// Mr. Joseph Fourier comes to the rescue.
//
FastFourierTransVisitor<double, DateTime> fft;

Expand All @@ -360,8 +336,8 @@ int main(int, char *[]) {
const auto &magnitudes = fft.get_magnitude();
double max_val = 0;

// The following analysis and conclusion are over simplified and naive. It is more involved
// which is behind the scope of Hello World. But this is the basic idea.
// The following analysis and conclusion are over simplified and naive. It is more involved which is behind the scope of
// Hello World. But this is the basic idea.
//
for (std::size_t i = 1; i < magnitudes.size(); ++i) {
const double val = 1.0 / magnitudes[i];
Expand Down

0 comments on commit bd918e4

Please sign in to comment.